You might think that a machine learning (ML) specialist company likeIntellegens is always pursuing the perfect model - one that takes a new set of system inputs and predicts their outputs correctly every time. But, despite the importance of model accuracy, it is possible to focus on it too much in real-world R&D.

A near-perfect model typically considered a model that predicts outputs reliably to within 5% - could mean thatmachine learning (ML)has found a set of robust relationships not previously observed by cutting through multi-dimensional complexity.

Image Credit: Intellegens Limited

However, this can also mean that experiments were poorly designed or trivial, and the ML is simply confirming the obvious. Such perfection is, in any case, mathematically unachievable in many complex systems with inherent uncertainties.

In the real world of R&D, a typical use case might be designing a set of experiments to find more effective formulations, chemicals, or materials. Here, visualizing the range of possibilities is beyond the capacity of the human brain and even relatively sophisticated Design of Experiments methods still result in large, expensive and time-consuming experimental programs. Users dont want perfection they just want ML to shift the odds in their favor, with predictions that outperform the logic currently driving their work.

Pursuing the ideal model may also waste time that is better spent elsewhere. It may also lead to users inadvertently narrowing down their search space in ways that exclude more innovative solutions.

Instead of asking how accurate a model is, the right question may focus on the models usefulness. Below are Intellegens top five examples of questions that might help a user to shape their model:

1. Can we get to an answer in fewer experiments?

Does the ML that is being used have the ability to understand what missing data could best improve its accuracy? This information can then be deployed to decide what experiment to perform next, resulting in a significantly reduced time-to-market. In some cases, theAlchemitesoftware from Intellegens has reduced experimental workloads by 80%+. More commonly, reductions of 50% are reported.

2. How do we generate new ideas for formulations that achieve our goals?

New concepts with a chance of success can result from a moderately-accurate model. And R&D teams are given a big helping hand if the model comes with a robust estimate of its uncertainty, pointing them towards those most likely to succeed. If the ML can move the dial so that one in three candidate formulations succeed when the previous metric was one in five, this could make a big difference.

3. Can we remove costly or environmentally harmful ingredients?

Questions like this typically derive from consumer, regulatory, or market pressure and require a fast response. ML can screen potential solutions, and an indication of probable success can be given by quantifying the uncertainty of the predictions.

4. Where should we focus which inputs are the most significant?

The absolute accuracy of predictions may be less important than whether useful relationships are identified, for example, between structure, processing variables, and properties. Often, the latter is the most vital piece of information that users need. A series ofanalytical toolsthat enable users to explore the sensitivity of outputs to particular inputs are provided by Alchemite.

5. Can we make better use of the expertise weve already developed?

Insight developed at great expense in R&D projects is often not be re-used. A valuable starting point for future projects can be provided by the ability to capture this insight in anML model.

Alchemite Analytics How Do Changes in Inputs Impacts Outputs?

Rather than focusing on ML as a magic bullet, it is essential to consider its use in informing scientific intuition and functioning alongside it.

Image Credit: Intellegens Limited

It is vital to have the right tools like uncertainty quantification and graphical analytics to interrogate and understand the results. When data is messy, as it often is in R&D, rather than investing up-front effort to clean and enrich the data, it can be valuable to be able to generate an ML model even an imperfect one quickly. By exploring this model, users can gain insight and improve their work iteratively and at a much lower cost.

The team at Intellegens values accurate models, and sometimes, they are, of course, essential. Mostly they also work in the spirit of the aphorism commonly attributed to statistician George Box:All models are wrong; some are useful.

This information has been sourced, reviewed and adapted from materials provided by Intellegens Limited.

For more information on this source, please visit Intellegens Limited.

Read more here:
Its Not Just About Accuracy - Five More things to Consider for a Machine Learning Model - AZoM

A total of 963 E. coli UTI patients from NCKUH were included, 14.2% of them had E. coli RUTI. All the 137 RUTI patients included in this study had RUTI caused by E. coli, 74 patients (54%) had 2 episodes of UTI within 6months and 63 patients (46%) had 3 episodes of UTI within 12months. All these episodes of E. coli related RUTI in this study were reinfection (recurrence of UTI with the same organisms in more than 2weeks). The duration of antibiotic treatment varied from 3 to 14days, and the antibiotic regimens included empirical antibiotic therapy and definitive antibiotic therapy according to the antimicrobial susceptibility test. The patient characteristics related to UTI and RUTI caused by E. coli are shown in Table 1. The median age was 67 and 75years for patients with UTI and RUTI, respectively. Compared to the UTI group, patients with RUTI had an older age, a greater prevalence of diabetes mellitus, liver cirrhosis, indwelling Foley catheter, neurogenic bladder, more frequent hospitalization/emergency department (ED) visit/UTI within 2years and any UTI symptom, and a worse renal function (Table 1).

The bacterial characteristic factors (phylogenicity, virulence genes, and antimicrobial susceptibility) related to UTI and RUTI are shown in Tables 2 and 3, respectively. Compared to those in the UTI group, E. coli isolates derived from the RUTI group had a lower prevalence of papG II, usp, ompT, and sat genes, and a higher prevalence of antimicrobial resistance in several antibiotics (including cefazolin, cefuroxime, cefixime, and levofloxacin).

The analysis results suggested RF model was better than the LR and DT model for RUTI prediction in the clinical visit. The 32 factors considered in the models for the first stage were age, gender, comorbidities (Dis1~Dis12), UTI symptoms (UTI_symptom1~UTI_symptom8), serum creatinine, frequency of hospitalization/emergency department (ED) visit/UTI within 2years (Pre_hos_2y, Pre_UTI_ER_2y, Pre_UTI_hos_2y), urinary red blood cell/HPF (URBC_level), urinary white blood cell (WBC)/high power field (HPF) (UWBC_level), urinary bacterial count (UBact), peak blood WBC count (BloodWBC), place (outpatient or ED) of urine sample collection (Place_of_collection), and disease group (four_disease_group). These factors are labeled in Table 1.

URBC_level and UWBC_level represent the rescaled level of the URBC and UWBC with values from 0 to 4 and from 1 to 4, respectively. The values 0, 1, 2, 3, and 4 of the URBC_level and UWBC_level corresponded to the ranges 0, 1~10, 11~100, 101~1000, and greater than 1000 per HPF, respectively. Place_of_collection indicates the place of urine sample collection, including outpatient clinic and ED. A new factor called four_disease_group was defined for RUTI prediction with value 0 or 1. We set four_disease_group value to 1 when one of the following diseases with anatomical or functional defect of urinary tract is present: indwelling Foley catheter (Dis5), obstructive uropathy (Dis6), urolithiasis (Dis7), and neurogenic bladder (Dis9). We would like to confirm the relation of four_disease_group with RUTI.

Regarding the validation results of fitted models to predict the development of RUTI in the clinical visit, Table 4 shows that the mean validation accuracy of RF is 0.700 which is higher than the results of LR and DT. The mean validation sensitivity and specificity of RF are 0.626 and 0.712, respectively. The standard deviations of estimated validation accuracy, sensibility, and specificity are 0.039, 0.131, and 0.046, respectively, which support the stability of RF model prediction. Note that the RUTI rate is only 136/963=0.138 which is relatively low for the observed samples. A nave model would predict non of the patients to have RUTI with a high accuracy 827/963=0.862. However, such prediction will lead to a very poor sensitivity with value 0. The RF model avoided such serious bias and provided a balance prediction capability in both sensitivity and specificity. The key technique in the RF model training is the usage of upsampling.

Variable importance in RF is evaluated by the mean decrease of accuracy in predictions on the out of bag samples when a given variable is excluded from the model. For example, if the age is taken away, the model prediction will reduce the accuracy rate by 11.9%. Figure1 is the variable importance plot of the RF analysis and shows that age, cirrhosis (Dis4), diabetes mellitus (Dis1), and disease group (four_disease_group) are the most important factors to predict recurrence of UTI in the clinical visit. Each of the 4 factors contributed around 10% prediction accuracy in the RF model.

Variable importance plot of the first stage RF analysis in percentage of mean decrease accuracy for the factors. It shows that age, cirrhosis (Dis4), diabetes mellitus (Dis1), and disease group (four_disease_group) are the most important 4 factors to predict recurrence in the clinical visit (sample size = 963).

A DT model is able to construct the decision rules for RUTI classification and provides the order of importance of the factors at the same time. Table 4 shows that the mean validation accuracy, sensitivity, and specificity of DT model are 0.654, 0.618, and 0.660, respectively. Although the validation accuracy of the DT is less than the values of the RF model, the results of DT model has its own edge in decision rule construction.

To obtain more insight on the RUTI factors in the clinical visit, one can check on Fig.2 which is the decision rules of the DT model built from all the 963 patients. The purpose of building a DT model with all collected data is to construct the decision rules for RUTI classification. In a DT model, when the patients satisfy the node's condition, the patients will be allocated to the left path of the node, otherwise the patients will be allocated to the right path of the node. The classification accuracy of this tree is 0.88, and the sensitivity and specificity are 0.26 and 0.98, respectively. Although the sensitivity is low due to the unbalanced rates of RUTI and UTI in the DT model, there are several valuable rules for RUTI classification. The 2 green boxes and 1 red box in Fig.2 indicate the nodes of the decision rules with a accuracy rate higher than 0.85 and 0.70 for non RUTI and RUTI classification, respectively. The three decision rules are:

When the factor states of a patient are without neurogenic bladder (Dis9=0) and without hospitalized within 2years (Pre_hos_2y<1), this rule claims that the patient will have no RUTI with classification accuracy 439/(439+34)=0.92.

When the factor states of a patient are without neurogenic bladder (Dis9=0), with previous hospitalization at least one time within 2years (Pre_hos_2y>=1), with serum creatinine less than 0.93mg/dL (creatinine<0.93), without cirrhosis (Dis4=0), and previous ER for UTI less than two times within 2years (Pre_UTI_ER_2y<2), this rule claims that the patient will have no RUTI with classification accuracy 296/(296+46)=0.86.

When the factor states of a patient are without neurogenic bladder (Dis9=0), with previous hospitalization at least one time within 2years (Pre_hos_2y>=1), with serum creatinine in the range between 0.74 and 3.9mg/dL (0.74

The decision rules of the DT analysis for development of RUTI in the clinical visit. (sample size = 963). The 2 green boxes and 1 red box indicate the nodes of the decision rules with an accuracy rate higher than 0.85 and 0.70 for non RUTI and RUTI classification, respectively.

The analysis results suggested RF model was better than the LR and DT model for RUTI prediction after hospitalization. The 62 factors considered in the models for the second stage not only contain the 32 factors used in the first stage analysis, but also include phylogenicity, 16 virulence genes, 11 antimicrobial susceptibility, Bacterial_Name, UTI_pos, Hospitalday, and Place_of_collection. The genes and antimicrobial are labeled in Table 2. Bacterial_name indicates Escherichia coli with or without extended spectrum -lactamase (ESBL). UTI_pos represents the location of urinary tract infection. Hospital_day gives the length (day) of hospital stay. Place_of_collection records the place of sample collection at ER, hospital, or outpatient clinic.

Regarding the validation results of refitted models to predict the development of RUTI after hospitalization, Table 5 shows that the mean validation accuracy of RF is 0.709 which is higher than the results of LR and DT. The mean validation sensitivity and specificity of RF are 0.620 and 0.722, respectively. The standard deviations of estimated validation accuracy, sensibility, and specificity are 0.047, 0.057, and 0.058, respectively, which support the stability of RF model prediction. Note that the RUTI rate is only 112/809=0.138 which is relatively low for the observed samples. A nave model would predict non of the patients to have RUTI with a high accuracy 697/809=0.862. However, such prediction will lead to a very poor sensitivity with value 0. The RF model avoided such serious bias and provided a balance prediction capability in both sensitivity and specificity.

Variable importance plot shows that based upon the mean decrease of accuracy in predictions on the out of bag samples when a given variable is excluded from the model. For example, if the cefixime (Anti7) is taken away, the model prediction will reduce the accuracy rate by 9.14%. Figure3 is the variable importance plot of the RF analysis and shows that cefixime (Anti7), afa (Gene11), usp (Gene8), and cefazolin (Anti5) are important factors to predict recurrence after hospitalization. Each of the 4 factors contributed more than 8% prediction accuracy in the RF model.

Variable importance plot of the second stage RF analysis in percentage of mean decrease accuracy for the factors. It shows that cefixime (Anti7), afa (Gene11), usp (Gene8), and cefazolin (Anti5) are important factors to predict recurrence after hospitalization (sample size = 809).

To obtain more insight on the RUTI factors after hospitalization, one can check on Fig.4 which is the decision rules of the DT model built from all the 803 patients. The classification accuracy of this tree is 0.89, and the sensitivity and specificity are 0.27 and 0.99, respectively. Although the sensitivity is low due to the unbalanced rates of RUTI and UTI in the DT model, there are several valuable rues for RUTI classification. The 4 green boxes and 3 red boxes in Fig.4 indicate the nodes of the decision rules with an accuracy rate higher than 0.85 and 0.70 for non RUTI and RUTI classification, respectively. The 7 decision rules are:

When the factor states of a patient are bacterial phylogenetic group B2 (Gene17=3) and the age less than 76years old (Age<76), this rule claims that the patient will have no RUTI with classification accuracy 322/(322+18)=0.94.

When the factor states of a patient are bacterial phylogenetic group B2 (Gene17=3), the age over 76years old (Age (ge) 76), and serum creatinine less than 3.5mg/dL (creatinine<3.5), this rule claims that the patient will have no RUTI with classification accuracy 148/(148+21)=0.87.

When the factor states of a patient are bacterial phylogenetic group B2 (Gene17=3), the age over 76years old (Age (ge) 76), serum creatinine less than 3.5mg/dL (creatinine (ge) 3.5), and more than 19days of hospital stay (Hospital_day (ge) 19), this rule claims that the patient will have RUTI with classification accuracy 8/(3+8)=0.72.

When the factor states of a patient are non-group B2 in bacterial phylogenicity (Gene17 (ne) 3) and S or I type in levofloxacin susceptibility (Anti25=1, 2), this rule claims that the patient will have no RUTI with classification accuracy 137/(137+22)=0.86.

When the factor states of a patient are non-group B2 in bacterial phylogenicity (Gene17 (ne) 3), R type in levofloxacin susceptibility (Anti25=3), bloodWBC more than 7.8 (bloodWBC (ge) 7.8), and group A or B1 in bacterial phylogenicity (Gene17=1, 2), this rule claims that the patient will have no RUTI with classification accuracy 42/(42+5)=0.89.

When the factor states of a patient are non-group B2 in bacterial phylogenicity (Gene17 (ne) 3), R type in levofloxacin susceptibility (Anti25=3), bloodWBC more than 7.8 (bloodWBC (ge) 7.8), group D in phylogenicity (Gene17=4), and more than 57days of hospital stay (Hospital_day (ge) 57), this rule claims that the patient will have RUTI with classification accuracy 6/(6+1)=0.85.

When the factor states of a patient are non-group B2 in bacterial phylogenicity (Gene17 (ne) 3), R type in levofloxacin susceptibility (Anti25=3), bloodWBC less than 7.8 (bloodWBC<7.8), and the value of UWBC more than 10 (UWBC_level (ne) 1), this rule claims that the patient will have RUTI with classification accuracy 16/(6+16)=0.72.

The decision rules of the DT analysis for development of RUTI after hospitalization. The 4 green boxes and 3 red boxes indicate the nodes of the decision rules with an accuracy rate higher than 0.85 and 0.70 for non RUTI and RUTI classification, respectively (sample size = 809).

Follow this link:
Machine learning to predict the development of recurrent urinary tract infection related to single uropathogen, Escherichia coli | Scientific Reports...

DUBLIN, Oct. 12, 2022 /PRNewswire/ --The "Global Markets for Machine Learning in the Life Sciences" report has been added to ResearchAndMarkets.com's offering.

Research and Markets Logo

This report highlights the current and future market potential for machine learning in life sciences and provides a detailed analysis of the competitive environment, regulatory scenario, drivers, restraints, opportunities and trends in the market. The report also covers market projections from 2022 through 2027 and profiles key market players.

The publisher analyzes each technology in detail, determines major players and current market status, and presents forecasts of growth over the next five years. Scientific challenges and advances, including the latest trends, are highlighted. Government regulations, major collaborations, recent patents and factors affecting the industry from a global perspective are examined.

Key machine learning in life sciences technologies and products are analyzed to determine present and future market status, and growth is forecast from 2022 to 2027. An in-depth discussion of strategic alliances, industry structures, competitive dynamics, patents and market driving forces is also provided.

Artificial intelligence (AI) is a term used to identify a scientific field that covers the creation of machines (e.g., robots) as well as computer hardware and software aimed at reproducing wholly or in part the intelligent behavior of human beings. AI is considered a branch of cognitive computing, a term that refers to systems able to learn, reason and interact with humans. Cognitive computing is a combination of computer science and cognitive science.

ML algorithms are designed to perform tasks such data browsing, extracting information that is relevant to the scope of the task, discovering rules that govern the data, making decisions and predictions, and accomplishing specific instructions. As an example, ML is used in image recognition to identify the content of an image after the machine has been instructed to learn the differences among many different categories of images.

There are several types of ML algorithms, the most common of which are nearest neighbor, naive Bayes, decision trees, a priori algorithms, linear regression, case-based reasoning, hidden Markov models, support vector machines (SVMs), clustering, and artificial neural networks. Artificial neural networks (ANN) have achieved great popularity in recent years for high-level computing.

Story continues

They are modeled to act similarly to the human brain. The most basic type of ANN is the feedforward network, which is formed by an input layer, a hidden layer and an output layer, with data moving in one direction from the input layer to the output layer, while being transformed in the hidden layer.

Report Includes

32 data tables and 28 additional tables

A comprehensive overview and up-to-date analysis of the global markets for machine learning in life sciences industry

Analyses of the global market trends, with historic market revenue data for 2020 and 2021, estimates for 2022, and projections of compound annual growth rates (CAGRs) through 2027

Highlights of the current and future market potential for ML in life sciences application, and areas of focus to forecast this market into various segments and sub-segments

Estimation of the actual market size for machine learning in life sciences in USD million values, and corresponding market share analysis based on solutions offering, mode of deployment, application, and geographic region

Updated information on key market drivers and opportunities, industry shifts and regulations, and other demographic factors that will influence this market demand in the coming years (2022-2027)

Discussion of the viable technology drivers through a holistic review of various platform technologies for new and existing applications of machine learning in the life sciences areas

Identification of the major stakeholders and analysis of the competitive landscape based on recent developments and segmental revenues

Emphasis on the major growth strategies adopted by leading players of the global machine learning in life sciences market, their product launches, key acquisitions, and competitive benchmarking

Profile descriptions of the leading market players, including Alteryx Inc., Canon Medical Systems Corp., Hewlett Packard Enterprise (HPE), KNIME AG, Microsoft Corp., and Phillips Healthcare

Key Topics Covered:

Chapter 1 Introduction

Chapter 2 Summary and Highlights

Chapter 3 Market Overview 3.1 Introduction 3.1.1 Understanding Artificial Intelligence in Healthcare 3.1.2 Artificial Intelligence in Healthcare Evolution and Transition

Chapter 4 Impact of the Covid-19 Pandemic 4.1 Introduction 4.1.1 Impact of Covid-19 on the Market

Chapter 5 Market Dynamics 5.1 Market Drivers 5.1.1 Investment in Ai Health Sector 5.1.2 Rising Chronic Diseases 5.1.3 Advanced, Precise Results 5.1.4 Increasing Research and Development Budget 5.2 Market Restraints and Challenges 5.2.1 Reluctance Among Medical Practitioners to Adopt Ai-Based Technologies 5.2.2 Privacy and Security of User Data 5.2.3 Hackers and Machine Learning 5.2.4 Ambiguous Regulatory Guidelines for Medical Software 5.3 Market Opportunities 5.3.1 Untapped Potential in Emerging Markets 5.4 Value Chain Analysis

Chapter 6 Market Breakdown by Offering 6.1 Software 6.1.1 Market Size and Forecast 6.2 Services 6.2.1 Market Size and Forecast

Chapter 7 Market Breakdown by Deployment Mode 7.1 Cloud 7.1.1 Market Size and Forecast 7.2 On-Premises 7.2.1 Market Size and Forecast

Chapter 8 Market Breakdown by Application 8.1 Diagnosis 8.1.1 Market Size and Forecast 8.2 Therapy 8.2.1 Market Size and Forecast 8.3 Healthcare Management 8.3.1 Market Size and Forecast

Chapter 9 Market Breakdown by Region 9.1 Global Market 9.2 North America 9.2.1 U.S. 9.2.1 Canada 9.3 Europe 9.3.1 Germany 9.3.2 U.K. 9.3.3 France 9.3.4 Italy 9.3.5 Spain 9.3.6 Rest of Europe 9.4 Asia-Pacific 9.4.1 China 9.4.2 Japan 9.4.3 India 9.4.4 Rest of Asia-Pacific 9.5 Rest of the World

Chapter 10 Regulations and Finance 10.1 Regulatory Framework 10.1.1 American Diabetes Association's Standards of Medical Care in Diabetes 10.1.2 Ata Guidelines for Artificial Intelligence 10.1.3 Indian Ai Guidelines, Strategy, and Standards

Chapter 11 Competitive Landscape 11.1 Overview 11.1.1 Development 11.1.2 Cloud 11.1.3 Users 11.1.4 Parent Market: Global Artificial Intelligence Market

Chapter 12 Company Profiles

For more information about this report visit https://www.researchandmarkets.com/r/qc8qjo

Media Contact:

Research and MarketsLaura Wood, Senior Managerpress@researchandmarkets.com

For E.S.T Office Hours Call +1-917-300-0470For U.S./CAN Toll Free Call +1-800-526-8630For GMT Office Hours Call +353-1-416-8900

U.S. Fax: 646-607-1904Fax (outside U.S.): +353-1-481-1716

Logo: https://mma.prnewswire.com/media/539438/Research_and_Markets_Logo.jpg

Cision

View original content:https://www.prnewswire.com/news-releases/outlook-on-the-machine-learning-in-life-sciences-global-market-to-2027---featuring-alteryx-anaconda-canon-medical-systems-and-imagen-technologies-among-others-301647240.html

SOURCE Research and Markets

View post:
Outlook on the Machine Learning in Life Sciences Global Market to 2027 - Featuring Alteryx, Anaconda, Canon Medical Systems and Imagen Technologies...

Why we write about this topic:

AI and Deep Learning are influencing our lives in a never before seen way. Albert van Breemen helps us understand the consequences.

Kivis algorithm is plain and simple: every second Friday of the month, together with Mikrocentrum, they invite a lecturer in the field of engineering for a talk about their field of expertise, followed by an opportunity to meet other engineers. Today, the floor in the AI Innovation Center at High tech Campus Eindhoven was for Albert van Breemen, CEO and CTO of VBTI, an AI engineering company that develops Deep Learning solutions for companies in agriculture and manufacturing. Van Breemen, whose company recently won a Gerard & Anton Award, guided his audience through the hidden tracks of Artificial Intelligence and deep learning.

VBTI has successfully applied deep learning technology to agricultural robots and harvest forecasting systems. This required the development of a dedicated platform to get deep learning operational: AutoDL. The platform has automated many of the lifecycle tasks of deep learning development; with the support of VDL, VBTI is now taking the technology to a new level.

VBTI introduces robots fitted with smart camera tech to agriculture

His work is all about making automation intelligent, Van Breemen says at the start of his lecture. We want to help industries like agriculture, manufacturing, logistics, and robotics in their transformation processes, using deep learning and computer vision. But first, what is deep learning?

Van Breemen presents a timeline showing three significant periods. Not many people are aware of it, but artificial intelligence was already mastered in the 1950s, by creating machines that could sense, reason, act, and adapt. In the 1980s, we had the second wave called Machine Learning. It was the age of the algorithms that used data to improve their performance. Neural networks were created. Only after 2005 can we speak of Deep Learning: we started training deep neural networks with big data.

Big data is crucial for this: the more data, the more deep learning options. But its not like humans are out of work because of this development, Van Breemen says. Most importantly, we need people to collect and select the data and annotate all this so the machine can actually learn from it. After these processes, the model training can start, and finally, its time for deployment.

Deep learning consumer successes can be found in autonomous driving, GO and chess achievements in GO and chess, or smart assistants like Siri or Alexa. And now, its time for the industrial domain to get into deep learning. VBTI/VDL is doing this by using AI to develop a cucumber leaf-cutting robot. Thats a really complex world because not one cucumber leaf or stem is the same, and still, the machine needs to recognize them exactly. All those variations make it difficult, the deep learning toolbox can make this process robust.

Van Breemen and his team have been working for years on the technology to support de de-leafing robot. Getting 80 percent accuracy is easy, but the last 20 percent is extremely difficult. You always wonder which and how much data should be collected and annotated, how you handle storage and versioning, and how you can tell what the quality of the data is. Van Breemen says he is happy and proud about the result, leading to an effective robot, but he also knows that this can never be the end of it. You never stop learning. You keep collecting new data more data is more deep learning capacity.

See the original post here:
The more data, the more deep learning capacity - Innovation Origins

DENVER & CHICAGO--(BUSINESS WIRE)--Forensic Discovery, a leader in digital forensic and eDiscovery services for the legal industry and corporations, announced that it is expanding its technology offering with Reveal, the global provider of the leading AI-powered eDiscovery and investigations platform. Reveal uses adaptive AI, behavioural analysis, and pre-trained AI model libraries to help uncover connections and patterns buried in large volumes of unstructured data.

Forensic Discovery is excited to offer next generation Artificial Intelligence to its hosted review and data analytics services through use of Reveal, said Trent Walton, founder of Forensic Discovery. Our clients, which range from the Am Law 100 to the Fortune 500, will greatly benefit from having the power to investigate, review and produce their data in new ways, thereby reducing litigation costs.

Forensic Discovery will leverage the platform globally to unlock intelligence that will help clients mitigate risks across a range of areas including litigation, investigations, compliance, ethics, fraud, human resources, privacy and security.

As we continue to expand the depth and breadth of Reveals marketplace offerings, we are excited to partner with Forensic Discovery, a demonstrated leader in digital forensics and eDiscovery, said Wendell Jisa, Reveals CEO. By taking full advantage of Reveals powerful platform, Forensic Discovery now has access to the industrys leading SaaS-based, AI-powered technology stack, helping them and their clients solve their most complex problems with greater intelligence.

For more information about Reveal-Brainspace and its AI platform for legal, enterprise and government organizations, visit http://www.revealdata.com.

About Forensic Discovery

Forensic Discovery is a litigation case management firm with expertise in Digital Forensics, eDiscovery, and Expert Testimony. The company has developed a proprietary workflow that allows its clients to forensically collect, filter, review, and produce electronic evidence using a hosted review platform. With offices in Colorado, California and Texas, Forensic Discovery is a leader in digital forensic and eDiscovery services for the legal industry and corporations. Learn more about the companys offerings by visiting http://www.forensicdiscovery.expert.

About Reveal

Reveal-Brainspace is a global provider of the leading AI-powered eDiscovery platform. Fuelled by powerful AI technology and backed by the most experienced team of data scientists in the industry, Reveals cloud-based software offers a full suite of eDiscovery solutions all on one seamless platform. Users of Reveal include law firms, Fortune 500 corporations, legal service providers, government agencies and financial institutions in more than 40 countries across five continents. Featuring deployment options in the cloud or on-premises, an intuitive user design and multilingual user interfaces, Reveal is modernizing the practice of law, saving users time and money and offering them a competitive advantage. For more information, visit http://www.revealdata.com.

Continue reading here:
Forensic Discovery Taps Reveal-Brainspace to Bolster its Analytics, AI and Machine Learning Capabilities - Business Wire

WHO releases country estimates on air pollution exposure and health impact, <https://www.who.int/news/item/27-09-2016-who-releases-country-estimates-on-air-pollution-exposure-and-health-impact> (2016).

Faridi, S. et al. Long-term trends and health impact of PM2.5 and O3 in Tehran, Iran, 20062015. Environ. Int. 114, 3749. https://doi.org/10.1016/j.envint.2018.02.026 (2018).

CAS PubMed Google Scholar

Sun, G. et al. Association between air pollution and the development of rheumatic disease: A systematic review. Int. J. Rheumatol. 2016, 111 (2016).

Google Scholar

Zhang, H. et al. Ambient air pollution exposure and gestational diabetes mellitus in Guangzhou, China: A prospective cohort study. Sci. Total Environ. 699, 134390. https://doi.org/10.1016/j.scitotenv.2019.134390 (2020).

ADS CAS PubMed Google Scholar

Rovira, J., Domingo, J. L. & Schuhmacher, M. Air quality, health impacts and burden of disease due to air pollution (PM10, PM2.5, NO2 and O3): Application of AirQ+ model to the Camp de Tarragona County Catalonia. Spain. Sci. Total Environ. 703, 135538. https://doi.org/10.1016/j.scitotenv.2019.135538 (2020).

ADS CAS PubMed Google Scholar

Mullen, C., Grineski, S. E., Collins, T. W. & Mendoza, D. L. Effects of PM2.5 on third grade students proficiency in math and english language arts. Int. J. Environ. Res. Public Health. 17, 6931. https://doi.org/10.3390/ijerph17186931 (2020).

PubMed Central Google Scholar

Delgado-Saborit, J. M. et al. A critical review of the epidemiological evidence of effects of air pollution on dementia, cognitive function and cognitive decline in adult population. Sci. Total Environ. 757, 143734 (2021).

ADS CAS PubMed Google Scholar

Peters, R. et al. Air pollution and dementia: A systematic review. J. Alzheimers Dis. 70, S145S163 (2019).

CAS PubMed PubMed Central Google Scholar

Shi, L. et al. A national cohort study (20002018) of long-term air pollution exposure and incident dementia in older adults in the United States. Nat. Commun. 12, 19 (2021).

ADS Google Scholar

Weuve, J. et al. Exposure to air pollution in relation to risk of dementia and related outcomes: An updated systematic review of the epidemiological literature. Environ. Health Perspect. 129, 096001 (2021).

CAS PubMed Central Google Scholar

Chen, J.-H. et al. Long-term exposure to air pollutants and cognitive function in taiwanese community-dwelling older adults: A four-year cohort study. J. Alzheimers Dis. 8, 115 (2020).

Google Scholar

Gao, Q. et al. Long-term ozone exposure and cognitive impairment among Chinese older adults: A cohort study. Environ. Int. 160, 107072 (2022).

CAS PubMed Google Scholar

He, F. et al. Impact of air pollution exposure on the risk of Alzheimers disease in China: A community-based cohort study. Environ. Res. 205, 112318 (2022).

CAS PubMed Google Scholar

Ran, J. et al. Long-term exposure to fine particulate matter and dementia incidence: A cohort study in Hong Kong. Environ. Pollut. 271, 116303 (2021).

CAS PubMed Google Scholar

Garcia, C. A., Yap, P.-S., Park, H.-Y. & Weller, B. L. Association of long-term PM2.5 exposure with mortality using different air pollution exposure models: Impacts in rural and urban California. Int J. Environ. Health Res. 26, 145157. https://doi.org/10.1080/09603123.2015.1061113 (2016).

CAS PubMed Google Scholar

Wang, B. et al. The impact of long-term PM2. 5 exposure on specific causes of death: exposure-response curves and effect modification among 53 million US Medicare beneficiaries. Environ. Health 19, 112 (2020).

CAS PubMed PubMed Central Google Scholar

Yu, W., Guo, Y., Shi, L. & Li, S. The association between long-term exposure to low-level PM2.5 and mortality in the state of Queensland, Australia: A modelling study with the difference-in-differences approach. PLOS Med. 17, e1003141. https://doi.org/10.1371/journal.pmed.1003141 (2020).

CAS PubMed PubMed Central Google Scholar

Bellinger, C., Jabbar, M. S. M., Zaane, O. & Osornio-Vargas, A. A systematic review of data mining and machine learning for air pollution epidemiology. BMC Public Health 17, 119 (2017).

Google Scholar

Belotti, J. T. et al. Air pollution epidemiology: A simplified Generalized Linear Model approach optimized by bio-inspired metaheuristics. Environ. Res. 191, 110106. https://doi.org/10.1016/j.envres.2020.110106 (2020).

CAS PubMed Google Scholar

Stingone, J. A., Pandey, O. P., Claudio, L. & Pandey, G. Using machine learning to identify air pollution exposure profiles associated with early cognitive skills among U.S. children. Environ. Pollut. 230, 730740. https://doi.org/10.1016/j.envpol.2017.07.023 (2017).

CAS PubMed PubMed Central Google Scholar

Chang, F.-J., Chang, L.-C., Kang, C.-C., Wang, Y.-S. & Huang, A. Explore spatio-temporal PM2.5 features in northern Taiwan using machine learning techniques. Sci. Total Environ. 736, 139656. https://doi.org/10.1016/j.scitotenv.2020.139656 (2020).

ADS CAS PubMed Google Scholar

Silibello, C. et al. Spatial-temporal prediction of ambient nitrogen dioxide and ozone levels over Italy using a random forest model for population exposure assessment. Air Qual. Atmos. Health 14, 817829. https://doi.org/10.1007/s11869-021-00981-4 (2021).

CAS Google Scholar

Fecho, K. et al. A novel approach for exposing and sharing clinical data: The translator integrated clinical and environmental exposures service. J. Am. Med. Inform. Assoc. 26, 10641073. https://doi.org/10.1093/jamia/ocz042 (2019).

PubMed PubMed Central Google Scholar

Chang, V., Ni, P. & Li, Y. K-clustering methods for investigating social-environmental and natural-environmental features based on air quality index. IT Prof. 22, 2834. https://doi.org/10.1109/MITP.2020.2993851 (2020).

Google Scholar

Wu, X., Cheng, C., Zurita-Milla, R. & Song, C. An overview of clustering methods for geo-referenced time series: From one-way clustering to co- and tri-clustering. Int. J. Geogr. Inf. Sci. 34, 18221848. https://doi.org/10.1080/13658816.2020.1726922 (2020).

Google Scholar

Karri, R., Chen, Y.-P.P. & Drummond, K. J. Using machine learning to predict health-related quality of life outcomes in patients with low grade glioma, meningioma, and acoustic neuroma. PLoS ONE 17, e0267931. https://doi.org/10.1371/journal.pone.0267931 (2022).

CAS PubMed PubMed Central Google Scholar

Hautamki, M. et al. The association between charlson comorbidity index and mortality in acute coronary syndromethe MADDEC study. Scand. Cardiovasc. J. 54, 146152. https://doi.org/10.1080/14017431.2019.1693615 (2020).

PubMed Google Scholar

Kantidakis, G. et al. Survival prediction models since liver transplantationcomparisons between Cox models and machine learning techniques. BMC Med. Res. Methodol. 20, 277. https://doi.org/10.1186/s12874-020-01153-1 (2020).

PubMed PubMed Central Google Scholar

Blom, M. C. et al. Training machine learning models to predict 30-day mortality in patients discharged from the emergency department: A retrospective, population-based registry study. BMJ Open 9, e028015. https://doi.org/10.1136/bmjopen-2018-028015 (2019).

PubMed PubMed Central Google Scholar

Weng, S. F., Reps, J., Kai, J., Garibaldi, J. M. & Qureshi, N. Can machine-learning improve cardiovascular risk prediction using routine clinical data?. PLoS ONE 12, e0174944. https://doi.org/10.1371/journal.pone.0174944 (2017).

CAS PubMed PubMed Central Google Scholar

Weng, S. F., Vaz, L., Qureshi, N. & Kai, J. Prediction of premature all-cause mortality: A prospective general population cohort study comparing machine-learning and standard epidemiological approaches. PLoS ONE 14, e0214365. https://doi.org/10.1371/journal.pone.0214365 (2019).

CAS PubMed PubMed Central Google Scholar

Chun, M. et al. Stroke risk prediction using machine learning: a prospective cohort study of 0.5 million Chinese adults. J. Am. Med. Inf. Assoc. 28, 17191727. https://doi.org/10.1093/jamia/ocab068 (2021).

Google Scholar

Moncada-Torres, A., van Maaren, M. C., Hendriks, M. P., Siesling, S. & Geleijnse, G. Explainable machine learning can outperform Cox regression predictions and provide insights in breast cancer survival. Sci. Rep. 11, 6968. https://doi.org/10.1038/s41598-021-86327-7 (2021).

ADS CAS PubMed PubMed Central Google Scholar

Du, M., Haag, D. G., Lynch, J. W. & Mittinty, M. N. Comparison of the tree-based machine learning algorithms to cox regression in predicting the survival of oral and pharyngeal cancers: Analyses based on SEER database. Cancers 12, 2802. https://doi.org/10.3390/cancers12102802 (2020).

CAS PubMed Central Google Scholar

Kim, H., Park, T., Jang, J. & Lee, S. Comparison of survival prediction models for pancreatic cancer: Cox model versus machine learning models. Genomics Inform. 20, e23. https://doi.org/10.5808/gi.22036 (2022).

PubMed PubMed Central Google Scholar

Kattan Michael, W. Comparison of Cox Regression with other methods for determining prediction models and nomograms. J. Urol. 170, S6S10. https://doi.org/10.1097/01.ju.0000094764.56269.2d (2003).

CAS PubMed Google Scholar

Lin, J., Li, K. & Luo, S. Functional survival forests for multivariate longitudinal outcomes: Dynamic prediction of Alzheimers disease progression. Stat. Methods Med. Res. 30, 99111 (2021).

MathSciNet PubMed Google Scholar

Facal, D. et al. Machine learning approaches to studying the role of cognitive reserve in conversion from mild cognitive impairment to dementia. Int. J. Geriatr. Psychiatry 34, 941949 (2019).

PubMed Google Scholar

Spooner, A. et al. A comparison of machine learning methods for survival analysis of high-dimensional clinical data for dementia prediction. Sci. Rep. 10, 110 (2020).

MathSciNet Google Scholar

Wang, J. et al. Random forest model in the diagnosis of dementia patients with normal mini-mental state examination scores. J. Personal. Med. 12, 37. https://doi.org/10.3390/jpm12010037 (2022).

Google Scholar

Pinheiro, L. I. C. C. et al. Application of data mining algorithms for dementia in people with HIV/AIDS. Comput. Math. Methods Med. 2021, 4602465. https://doi.org/10.1155/2021/4602465 (2021).

PubMed PubMed Central Google Scholar

Brickell, E., Whitford, A., Boettcher, A., Pereira, C. & Sawyer, R. J. A-1 the influence of base rate and sample size on performance of a random forest classifier for dementia prediction: Implications for recruitment. Arch. Clin. Neuropsychol. 36, 10401040. https://doi.org/10.1093/arclin/acab062.19 (2021).

Google Scholar

Dauwan, M. et al. Random forest to differentiate dementia with Lewy bodies from Alzheimers disease. Alzheimers Dement. Diagn. Assess. Dis. Monit. 4, 99106. https://doi.org/10.1016/j.dadm.2016.07.003 (2016).

Google Scholar

Mar, J. et al. Validation of random forest machine learning models to predict dementia-related neuropsychiatric symptoms in real-world data. J. Alzheimers Dis. 77, 855864. https://doi.org/10.3233/JAD-200345 (2020).

PubMed PubMed Central Google Scholar

World Medical Association. World medical association declaration of Helsinki: Ethical principles for medical research involving human subjects. JAMA 310, 21912194. https://doi.org/10.1001/jama.2013.281053 (2013).

CAS Google Scholar

Taiwan Environmental Protection Administration (EPA) website, <https://airtw.epa.gov.tw/CHT/Query/His_Data.aspx>

Yu, H.-L. et al. Interactive spatiotemporal modelling of health systems: The SEKSGUI framework. Stoch. Env. Res. Risk Assess. 21, 555572. https://doi.org/10.1007/s00477-007-0135-0 (2007).

MathSciNet Google Scholar

Charlson, M. E., Pompei, P., Ales, K. L. & MacKenzie, C. R. A new method of classifying prognostic comorbidity in longitudinal studies: Development and validation. J. Chronic Dis. 40, 373383. https://doi.org/10.1016/0021-9681(87)90171-8 (1987).

CAS PubMed Google Scholar

Hude, Q. et al. Coding algorithms for defining comorbidities in ICD-9-CM and ICD-10 administrative data. Med. Care 43, 11301139 (2005).

Google Scholar

Heagerty, P. J. & Saha, P. SurvivalROC: Time-dependent ROC curve estimation from censored survival data. Biometrics 56, 337344 (2000).

CAS PubMed Google Scholar

Harrell Jr, F. E., Harrell Jr, M. F. E. & Hmisc, D. Package rms. Vanderbilt University, 229 (2017).

Harrell, F. E. Jr., Califf, R. M., Pryor, D. B., Lee, K. L. & Rosati, R. A. Evaluating the yield of medical tests. JAMA 247, 25432546. https://doi.org/10.1001/jama.1982.03320430047030 (1982).

Read this article:
Long-term exposure to particulate matter was associated with increased dementia risk using both traditional approaches and novel machine learning...

Oracle Machine Learning AutoML User Interface

A no-code user interface supporting AutoML on Oracle Autonomous Database to improve both data scientist productivity and non-expert user access to powerful in-database algorithms for classification and regression.

Accelerate machine learning modeling using Oracle Autonomous Database as a high performance computing platform with an R interface. Use Oracle Machine Learning Notebooks with R, Python, and SQL interpreters to develop machine learningbased solutions. Easily deploy user-defined R functions from SQL and REST APIs with data-parallel and task-parallel options.

Data scientists and other Python users accelerate machine learning modeling and solution deployment by using Oracle Autonomous Database as a high-performance computing platform with a Python interface. Built-in automated machine learning (AutoML) recommends relevant algorithms and features for each model, and performs automated model tuning. Together, these capabilities enhance user productivity, model accuracy, and scalability.

Data scientists and data analysts can use this drag-and-drop user interface to quickly build analytical workflows. Rapid model development and refinement allows users to discover hidden patterns, relationships, and insights in their data.

Read the rest here:
Machine Learning in Oracle Database | Oracle

Stay organized with collections Save and categorize content based on your preferences. Foundational courses

The foundational courses cover machine learning fundamentals and core concepts.

We recommend taking them in the order below.

New

A course to help you map real-world problems to machine learning solutions.

The advanced courses teach tools and techniques for solving a variety of machine learning problems.

The courses are structured independently. Take them based on interest or problem domain.

Clustering is a key unsupervised machine learning strategy to associate related items.

Our guides offer simple step-by-step walkthroughs for solving common machine learning problems using best practices.

Become a better machine learning engineer by following these machine learning best practices used at Google.

This guide assists UXers, PMs, and developers in collaboratively working through AI design topics and questions.

This comprehensive guide provides a walkthrough to solving text classification problems using machine learning.

This guide describes the tricks that an expert data analyst uses to evaluate huge data sets in machine learning problems.

The glossary defines general machine learning terms.

[{ "type": "thumb-down", "id": "missingTheInformationINeed", "label":"Missing the information I need" },{ "type": "thumb-down", "id": "tooComplicatedTooManySteps", "label":"Too complicated / too many steps" },{ "type": "thumb-down", "id": "outOfDate", "label":"Out of date" },{ "type": "thumb-down", "id": "samplesCodeIssue", "label":"Samples / code issue" },{ "type": "thumb-down", "id": "otherDown", "label":"Other" }] [{ "type": "thumb-up", "id": "easyToUnderstand", "label":"Easy to understand" },{ "type": "thumb-up", "id": "solvedMyProblem", "label":"Solved my problem" },{ "type": "thumb-up", "id": "otherUp", "label":"Other" }]

Read more:
Machine Learning | Google Developers

Microcontrollers, miniature computers that can run simple commands, are the basis for billions of connected devices, from internet-of-things (IoT) devices to sensors in automobiles. But cheap, low-power microcontrollers have extremely limited memory and no operating system, making it challenging to train artificial intelligence models on edge devices that work independently from central computing resources.

Training a machine-learning model on an intelligent edge device allows it to adapt to new data and make better predictions. For instance, training a model on a smart keyboard could enable the keyboard to continually learn from the users writing. However, the training process requires so much memory that it is typically done using powerful computers at a data center, before the model is deployed on a device. This is more costly and raises privacy issues since user data must be sent to a central server.

To address this problem, researchers at MIT and the MIT-IBM Watson AI Lab developed a new technique that enables on-device training using less than a quarter of a megabyte of memory. Other training solutions designed for connected devices can use more than 500 megabytes of memory, greatly exceeding the 256-kilobyte capacity of most microcontrollers (there are 1,024 kilobytes in one megabyte).

The intelligent algorithms and framework the researchers developed reduce the amount of computation required to train a model, which makes the process faster and more memory efficient. Their technique can be used to train a machine-learning model on a microcontroller in a matter of minutes.

This technique also preserves privacy by keeping data on the device, which could be especially beneficial when data are sensitive, such as in medical applications. It also could enable customization of a model based on the needs of users. Moreover, the framework preserves or improves the accuracy of the model when compared to other training approaches.

Our study enables IoT devices to not only perform inference but also continuously update the AI models to newly collected data, paving the way for lifelong on-device learning. The low resource utilization makes deep learning more accessible and can have a broader reach, especially for low-power edge devices, says Song Han, an associate professor in the Department of Electrical Engineering and Computer Science (EECS), a member of the MIT-IBM Watson AI Lab, and senior author of the paper describing this innovation.

Joining Han on the paper are co-lead authors and EECS PhD students Ji Lin and Ligeng Zhu, as well as MIT postdocs Wei-Ming Chen and Wei-Chen Wang, and Chuang Gan, a principal research staff member at the MIT-IBM Watson AI Lab. The research will be presented at the Conference on Neural Information Processing Systems.

Han and his team previously addressed the memory and computational bottlenecks that exist when trying to run machine-learning models on tiny edge devices, as part of their TinyML initiative.

Lightweight training

A common type of machine-learning model is known as a neural network. Loosely based on the human brain, these models contain layers of interconnected nodes, or neurons, that process data to complete a task, such as recognizing people in photos. The model must be trained first, which involves showing it millions of examples so it can learn the task. As it learns, the model increases or decreases the strength of the connections between neurons, which are known as weights.

The model may undergo hundreds of updates as it learns, and the intermediate activations must be stored during each round. In a neural network, activation is the middle layers intermediate results. Because there may be millions of weights and activations, training a model requires much more memory than running a pre-trained model, Han explains.

Han and his collaborators employed two algorithmic solutions to make the training process more efficient and less memory-intensive. The first, known as sparse update, uses an algorithm that identifies the most important weights to update at each round of training. The algorithm starts freezing the weights one at a time until it sees the accuracy dip to a set threshold, then it stops. The remaining weights are updated, while the activations corresponding to the frozen weights dont need to be stored in memory.

Updating the whole model is very expensive because there are a lot of activations, so people tend to update only the last layer, but as you can imagine, this hurts the accuracy. For our method, we selectively update those important weights and make sure the accuracy is fully preserved, Han says.

Their second solution involves quantized training and simplifying the weights, which are typically 32 bits. An algorithm rounds the weights so they are only eight bits, through a process known as quantization, which cuts the amount of memory for both training and inference. Inference is the process of applying a model to a dataset and generating a prediction. Then the algorithm applies a technique called quantization-aware scaling (QAS), which acts like a multiplier to adjust the ratio between weight and gradient, to avoid any drop in accuracy that may come from quantized training.

The researchers developed a system, called a tiny training engine, that can run these algorithmic innovations on a simple microcontroller that lacks an operating system. This system changes the order of steps in the training process so more work is completed in the compilation stage, before the model is deployed on the edge device.

We push a lot of the computation, such as auto-differentiation and graph optimization, to compile time. We also aggressively prune the redundant operators to support sparse updates. Once at runtime, we have much less workload to do on the device, Han explains.

A successful speedup

Their optimization only required 157 kilobytes of memory to train a machine-learning model on a microcontroller, whereas other techniques designed for lightweight training would still need between 300 and 600 megabytes.

They tested their framework by training a computer vision model to detect people in images. After only 10 minutes of training, it learned to complete the task successfully. Their method was able to train a model more than 20 times faster than other approaches.

Now that they have demonstrated the success of these techniques for computer vision models, the researchers want to apply them to language models and different types of data, such as time-series data. At the same time, they want to use what theyve learned to shrink the size of larger models without sacrificing accuracy, which could help reduce the carbon footprint of training large-scale machine-learning models.

AI model adaptation/training on a device, especially on embedded controllers, is an open challenge. This research from MIT has not only successfully demonstrated the capabilities, but also opened up new possibilities for privacy-preserving device personalization in real-time, says Nilesh Jain, a principal engineer at Intel who was not involved with this work. Innovations in the publication have broader applicability and will ignite new systems-algorithm co-design research.

On-device learning is the next major advance we are working toward for the connected intelligent edge. Professor Song Hans group has shown great progress in demonstrating the effectiveness of edge devices for training, adds Jilei Hou, vice president and head of AI research at Qualcomm. Qualcomm has awarded his team an Innovation Fellowship for further innovation and advancement in this area.

This work is funded by the National Science Foundation, the MIT-IBM Watson AI Lab, the MIT AI Hardware Program, Amazon, Intel, Qualcomm, Ford Motor Company, and Google.

More:
Learning on the edge | MIT News | Massachusetts Institute of Technology - MIT News

A review of studies published in JAMA Network Open found few randomized clinical trials for medical machine learning algorithms, and researchers noted quality issues in many published trials they analyzed.

The review included 41 RCTs of machine learning interventions. It found 39% were published just last year, and more than half were conducted at single sites. Fifteen trials took place in the U.S., while 13 were conducted in China. Six studies were conducted in multiple countries.

Only 11 trials collected race and ethnicity data. Of those, a median of 21% of participants belonged to underrepresented minority groups.

None of the trials fully adhered to the Consolidated Standards of Reporting Trials Artificial Intelligence (CONSORT-AI), a set of guidelines developed for clinical trials evaluating medical interventions that include AI. Thirteen trials met at least eight of the 11 CONSORT-AI criteria.

Researchers noted some common reasons trials didn't meet these standards, including not assessing poor quality or unavailable input data, not analyzing performance errors and not including information about code or algorithm availability.

Using the Cochrane Risk of Bias tool for assessing potential bias in RCTs, the study also found overall risk of bias was high in the seven of the clinical trials.

"This systematic review found that despite the large number of medical machine learning-based algorithms in development, few RCTs for these technologies have been conducted. Among published RCTs, there was high variability in adherence to reporting standards and risk of bias and a lack of participants from underrepresented minority groups. These findings merit attention and should be considered in future RCT design and reporting," the study's authors wrote.

WHY IT MATTERS

The researchers said there were some limitations to their review. They looked at studies evaluating a machine learning tool that directly impacted clinical decision-makingso future research could look at a broader range of interventions, like those for workflow efficiency or patient stratification. The review also only assessed studies through October 2021, and more reviews would be necessary as new machine learning interventions are developed and studied.

However, the study's authors said their review demonstrated more high-quality RCTs of healthcare machine learning algorithms need to be conducted. Whilehundreds of machine-learning enabled devices have been approved by the FDA, the review suggests the vast majority didn't include an RCT.

"It is not practical to formally assess every potential iteration of a new technology through an RCT (eg, a machine learning algorithm used in a hospital system and then used for the same clinical scenario in another geographic location)," the researchers wrote.

"A baseline RCT of an intervention's efficacy would help to establish whether a new tool provides clinical utility and value. This baseline assessment could be followed by retrospective or prospective external validation studies to demonstrate how an intervention'sefficacy generalizes over time and across clinical settings."

Read the original here:
Study: Few randomized clinical trials have been conducted for healthcare machine learning tools - Mobihealth News