Skip to content

Let's Data ScienceLEARN • BUILD • STAY AHEAD

News
Blog
Code Problems
Pricing
Contact

© 2026 Let's Data Science

Advertise|Terms|Privacy||Image Rights

Home
News
Machine Learning

Topic24 articles · Updated June 6, 2026

Machine Learning News

Machine learning coverage for working practitioners: research papers worth reading, framework and library updates, MLOps tooling, and applied ML stories shipping to production.

electronic-health-records
6.9
Jun 6, 2026
Researchers detect hidden self-harm histories with ML
The University of New Mexico School of Medicine led a study analyzing electronic health records for more than 1.3 million patients served by the Veterans Health Administration, according to a UNM press release. The researchers reported that diagnosis codes captured 1.85% of patients with documented self-harm history, while their machine learning method estimated documented self-harm in 7.9% of patients, over four times higher, per the study published in the Journal of Medical Internet Research (reported by News-Medical and UNM). The team combined a novel machine learning approach with expert chart review and statistical calibration, the UNM release states. The study also found that among veterans with a diagnosis code for self-harm, only 22.6% had self-harm listed on the VHA problem list. "For research and planning, if we only count what is easy to see in diagnosis codes, we may substantially underestimate the need for mental health services," said Christophe Lambert, PhD, per the UNM release.
Quick Read
cybersecurity
4.8
Jun 5, 2026
AI Strengthens Cybersecurity Detection and Response
CodeCondo published an overview of how artificial intelligence is applied in cybersecurity to detect, analyze, and respond to threats in real time, moving defenses beyond signature-based rules. The article lists core building blocks, including machine learning, deep learning, natural language processing, predictive analytics, behavioral analytics, and automated threat intelligence, and notes adoption across finance, healthcare, government, and e-commerce, per CodeCondo. It is a general explainer rather than a report of a specific product, incident, or research result. For practitioners, the useful framing is that automation and data-driven detection shift work from manual triage toward model-driven prioritization, while raising new demands around data quality, model maintenance, and adversarial robustness.
Quick Read
gastrointestinal-cancer
5.6
Jun 4, 2026
Study Advances GI Cancer Risk Prediction with ML
A peer-reviewed study published in JMIR Medical Informatics evaluates machine learning methods for predicting gastrointestinal (GI) cancer risk in a South Korean prospective cohort, where GI cancers are a major health burden. Analyzing 7,652 participants with 156 incident GI cancer cases (about 2%) over 14 years of follow-up, the authors tackle the severe class imbalance that makes rare-disease prediction difficult. They introduce a patient-centered undersampling technique (PCUSTe) modeled on frequency-matched case-control design and benchmark it against SMOTE, ADASYN, and hybrid resampling. An incrementally trained stochastic gradient descent model on PCUSTe data reached a sensitivity of 0.77 and an AUC of 0.77, while logistic regression without resampling produced balanced results (sensitivity 0.70, specificity 0.71, AUC 0.75). The authors frame the models as tools for earlier risk stratification and targeted screening.
Quick Read
proton
5.6
Jun 4, 2026
Proton deploys ML to detect service abuse
According to Infosecurity Magazine (as indexed by itsecuritynews.info), Proton uses machine learning models to detect abuse of its services, with a particular focus on email addresses exploited by cybercriminals. The sourced item is a short conference-report style piece tied to Infosecurity Europe and does not provide technical specifications, model names, or quoted commentary from Proton. Editorial analysis: Companies tracking similar abuse patterns typically combine automated detectors with human review to balance false positives and privacy constraints, and practitioners should treat the report as an example of operational ML applications in email and account-security contexts.
Quick Read
wearables
6.5
Jun 3, 2026
Researchers Develop Compact Wearable for Continuous Blood Pressure
A team from the University of Utah and the University of Illinois Chicago, with collaborators at Harvard Medical School and the University of Pittsburgh, built a cuffless, smartwatch-style wearable that measures blood pressure continuously, the University of Utah reports. The device passes an imperceptible electrical current through the wrist and uses bioimpedance - how easily current flows through blood and tissue - to capture changes with each heartbeat, then reconstructs a full blood-pressure waveform with a physics-informed machine learning model. Embedding the physics of blood flow keeps the model from predicting physically impossible values and, the team says, removes the need for per-user calibration. Graduate students tested it on 150 people, including intensive-care and outpatient patients. The work is set to appear in Nature Communications as Cuffless hemodynamic monitoring with physics-informed machine learning models, with an early version posted May 14.
Quick Read+1
epidemiology-modeling
6.0
Jun 3, 2026
Authors Apply ML to Hyperparameter Tuning in Pandemic Models
According to the arXiv preprint arXiv:2606.02650, "Using Machine Learning to Enhance Hyperparameter Optimization in Pandemic Modeling: Case study of COVID-19 Dynamics in Ghana," the authors integrate machine learning into hyperparameter estimation for compartmental epidemic models. The paper reformulates five country-specific COVID-19 models into a shared non-autonomous ordinary-differential-equation structure, applies Modified Patankar-Runge-Kutta (MPRK) schemes for positive, conservative numerical integration, embeds the solver in a cost function to fit piecewise-constant time-varying parameters, and uses a WENO reconstruction in post-processing to recover smoother coefficients. Per the abstract, the method achieves 5-day predictions within a 10% error range for the Ghana case study. The contribution is a structure-preserving workflow intended to keep short-term epidemic forecasts physically consistent.
Quick Read
machine-learning
6.4
May 27, 2026
Machine Learning Addresses Drug-Safety Evidence Gaps in Pregnancy
The JMIR News and Perspectives article by Michelle Falci reports that women, and especially pregnant women, have been systematically excluded from clinical research following historical FDA guidance and related policies. According to JMIR, a survey found that 79% of women take at least one medication during pregnancy, with 63.2% making at least one medication change and 36.5% modifying or stopping medication without clinician input. The article notes global efforts such as a World Health Organization task force aiming for ethical inclusion of pregnant and breastfeeding women in clinical research by 2030, and it highlights the FDA Pregnancy and Lactation Labeling Rule (PLLR) as an incomplete but recent step. JMIR reports researchers are turning to large datasets and machine learning, with an emphasis on interpretability and causal inference, to help close these evidence gaps.
Quick Read
wealth-management
6.9
May 27, 2026
Wealth Management Firms Adopt Data-Driven Personalization at Scale
HedgeCo.Net reports that wealth management is undergoing a structural transformation centered on data. According to HedgeCo.Net, advisory firms are moving beyond basic digital tools toward models that combine data science, artificial intelligence, automation, and portfolio engineering to deliver more precise, responsive, and personalized services. HedgeCo.Net describes practical applications including machine learning to detect client needs earlier, automated tax-aware portfolio adjustments, integrated financial-planning and behavioral signals, and unified client views that include private market exposure. Editorial analysis: For practitioners, this trend raises demand for data integration, model explainability, and production-grade risk systems rather than isolated analytics proofs of concept.
Quick Read
materials-science
6.8
May 26, 2026
Texas A&M develops closed-loop AI alloy discovery workflow
Researchers at Texas A&M University developed an AI-guided closed-loop workflow that combines computation, high-throughput thermodynamic modeling, experiments, and machine learning to accelerate discovery of high-temperature, heat-resistant alloys, the article reports. The workflow uses computational screening to winnow thousands of candidate compositions before selecting promising alloys for real-world experiments, which researchers said reduced the need for costly experimental testing, according to Interesting Engineering. The article includes a quote from doctoral student Cafer Acemi, who received the Acta Student Award for the campaign, and notes the experiment aimed to ensure candidate alloys could be manufactured at scale, the article reports.
Quick Read
black-holes
5.2
May 25, 2026
Author Uses Machine Learning to Constrain Black Hole Parameters
According to the arXiv abstract for arXiv:2605.22862, Maryem Jemri submitted a paper on 19 May 2026 titled "Constraining Black Hole Parameters in Non-Commutative Geometry using Machine Learning." The abstract reports the use of CUDA-based high-performance computations to study event horizon structure, shadow properties, and the energy emission rate of rotating and charged black holes in a non-commutative geometry framework with string clouds and dark energy sectors. The author builds numerical datasets and trains a fully connected neural network to compare model outputs with observational data from the Event Horizon Telescope collaborations. The abstract states, "we find that the non-commutative model under study is consistent with the observations of black holes."
Quick Read
machine-learning
6.1
May 19, 2026
Study develops ML dyslipidemia prediction across three cohorts
The study develops a machine learning prediction model for dyslipidemia and evaluates its association with atherothrombotic events. Evaluation uses three independent cohorts from South Korea, Japan, and the United Kingdom, focusing on algorithm development and cross-cohort assessment.
Quick Read
neuroimaging
6.6
May 19, 2026
AI analyzes brain scans to predict depression
Per National Elf Service's write-up of Jiang et al. (2026), researchers applied machine-learning and deep-learning methods to two brain-imaging datasets to search for brain-based markers of major depressive disorder. The first dataset was the UK Biobank, containing 1,496 MDD cases and 27,741 controls, with a four-to-one control matching and training/testing splits, according to the article. The team divided grey-matter images into 3D voxels and trained models to detect subtle, localized patterns. The National Elf Service characterises the study's results as modest and exploratory, noting prior large consortia such as ENIGMA MDD found structural changes but limited predictive power. The article frames the work as incremental evidence that newer AI methods may find subtler signals, while substantial translational challenges remain.
Quick Read
fall-prediction
6.8
May 14, 2026
Systematic Review Evaluates ML Models Predicting Future Falls
According to the JMIR preprint by Gao et al. (2026), the authors screened 6,865 records and included 27 longitudinal studies that developed or validated machine learning (ML) or deep learning (DL) models to predict future falls in community-dwelling adults aged >=60. Per the paper, 22 studies used ML and 5 used DL; prediction horizons ranged from 3 months to 4 years and reported fall incidence varied from 1.6% to 46.6%. Gao et al. searched major databases through August 31, 2025, excluded real-time detection, simulated/no-fall and inpatient studies, assessed risk of bias with PROBAST, and meta-analyzed logit-transformed AUCs using the Hartung-Knapp-Sidik-Jonkman (HKSJ) random-effects method with 95% prediction intervals estimated by four approaches. The authors also ran leave-one-out robustness checks and examined small-study effects with funnel plots and Egger-type tests.
Quick Read
medical-device
5.6
May 14, 2026
Study develops spectral ML for smart catheters
A proof-of-concept spectral-driven machine learning algorithm was developed and validated using smart urinary and drainage catheter systems. The paper frames the work as "From Flow to Feature" and notes current catheter systems collect fluids for visual inspection or manual sampling, offering limited diagnostic value.
Quick Read
fashion-technology
5.7
May 12, 2026
Synflux Wins $200K Trailblazer Prize for Waste Reduction
Synflux, a Japanese startup, was revealed as the winner of the Trailblazer Programme 2026 at the Global Fashion Summit in Copenhagen, according to Global Fashion Agenda. The company uses a proprietary system that combines machine learning and 3D simulation to optimise garment cutting patterns, a capability Global Fashion Agenda describes in its Trailblazer profile. At the event, Synflux CEO Kazuya Kawasaki said at least 30 percent of material is typically lost on the cutting-room floor and that Synflux's approach can yield "up to 66 percent less waste," per reporting by WWD. WWD also reports Synflux ran a pilot with The North Face that halved waste for one bestselling jacket. WWD states Synflux is poised to receive a $200,000 award from a $50 million PDS Ventures fund, pending final due diligence and committee approval; the prize includes strategic support from PDS Group and Positive Materials.
Quick Read
suicide-prevention
6.3
May 11, 2026
ML and Screening Improve Suicide Risk Detection for American Indian Patients
A retrospective study by Goklish et al., published in J Med Internet Res (2026) and indexed on PubMed, evaluated combining an electronic health record (EHR) based machine learning model with the Ask Suicide-Screening Questions (ASQ) at an Indian Health Service (IHS) Emergency Department. The cohort covered adult ED visits from October 1, 2019 to October 2, 2021, and performance was assessed using 10 cross-validated folds, per the paper. The authors operationalized a logistic regression model at 95th (high-risk) and 75th (medium-risk) percentiles. According to the J Med Internet Res article and a JoVE summary, the ML medium-risk threshold alone showed sensitivity 0.782, parallel testing (ML medium-risk alongside ASQ) achieved sensitivity 0.795 while capturing cases missed by ASQ, and a serial strategy (ML high-risk then ASQ) produced the highest positive predictive value (PPV 0.050) but reduced sensitivity.
Quick Read+1
zalando
4.1
May 10, 2026
Zalando Hosts Workshop on Optimization and ML
Per Zalando's engineering blog, Zalando hosted the 106th meeting of the GOR working group "Praxis der Mathematischen Optimierung" on October 6-7, 2022 at Zalando Headquarters in Berlin (GOR invitation PDF). The hybrid workshop brought roughly sixty in-person participants and streamed virtually, and included academic and industry talks covering forecasting, network design, pricing, logistics, scheduling, and vehicle routing (Zalando engineering blog). Zalando presenters included two internal use-cases: a pricing-team talk on large-scale article discounting and a logistics-team talk on stock distribution (Zalando engineering blog). The GOR invitation describes the meeting as focused on applied methods from mathematical optimization and machine learning for e-commerce (GOR invitation PDF).
Quick Read
sleep-apnea
6.0
May 8, 2026
Researchers Validate ML Multidimensional Oximetry for OSA Screening
Per a JMIR Medical Informatics preprint (2025), the paper titled "Machine Learning-Based Multidimensional Oximetry for Obstructive Sleep Apnea Screening: Development and External Validation" develops and externally validates a machine-learning approach that uses multidimensional oximetry signals for screening obstructive sleep apnea (OSA). The article's background notes that OSA affects nearly one billion people globally. The preprint reports model development and an external validation step, positioning the work as an attempt to evaluate generalizability of oximetry-based screening outside the development cohort. The available source is the JMIR preprint citation; the original paper text and performance metrics were not provided in the scraped materials.
Quick Read
telehealth
5.4
May 7, 2026
Researchers analyze telehealth resource use across specialties
Researchers analyze resource use patterns in US telehealth services across four specialties using machine learning and clustering methods. The study situates its analysis in the context of telehealth's expansion during the COVID-19 pandemic and the transformation of U.S. health care delivery.
Quick Read
eeg
6.0
Apr 30, 2026
Resting EEG Complexity Predicts but Does Not Generalize for iTBS Response
According to a preprint posted on Research Square and indexed in PubMed, Matthew Ning et al. evaluate whether baseline resting-state EEG complexity and transcranial magnetic stimulation measures can predict neurophysiological responses to a single session of intermittent theta-burst stimulation (iTBS) applied to primary motor cortex (Res Sq preprint, 2025 Sep 24; PubMed PMID 41041549). The authors used statistical reliability analyses and supervised machine-learning models that integrated baseline rsEEG features and TMS-evoked measures including motor-evoked potentials (MEPs) and TMS-evoked potentials (TEPs). The preprint reports that while some predictive signal was present within-sample, the models failed to generalize to independent validation, as reflected in the manuscript title and reported results in the preprint. The work is a preprint and has not been peer reviewed (PubMed).
Quick Read+1
cybersecurity
6.6
Apr 30, 2026
Australia Post partners with Alpha Level for cybersecurity
Australia Post has partnered with US-based AI security firm Alpha Level to apply machine learning to cyber threat detection across its national network. Reporting by ARNnet, Australian Cybersecurity Magazine and Post & Parcel describes the collaboration as using AI to collect, process and analyse billions of data points each month, with Australia Post's Chief Information Security Officer Adam Cartwright saying the organisation generates about four billion data points monthly. Cartwright is quoted saying machine learning helps models understand what "good" looks like and speeds up detection. Alpha Level co-founder and CTO Dr Josh Neil, a former Microsoft Principal Data Scientist, is quoted saying the work aims to "elevate threat detection speed and precision." Sources say the deployment will focus on reducing alert noise and isolating higher-confidence threats for analysts to review.
Quick Read+1
product-onboarding
5.2
Apr 26, 2026
Zalando Develops Content Creation Copilot for Onboarding
Per a Zalando engineering blog post published on Sep 18, 2024, the company describes an AI-assisted "Content Creation Copilot" designed to speed product onboarding by extracting product attributes and supporting quality assurance. The post says the manual content-creation stage contributed approximately 25% of the overall content production timeline and that copywriters currently perform attribute enrichment and QA using a Content Creation Tool under a four-eyes principle. According to the post, the Copilot provides assistive functions to detect and correct defects earlier in the workflow and to increase attribute coverage in the product catalog. Editorial analysis: Companies applying similar assistive ML for attribute extraction often reduce manual review time and improve catalogue completeness, which can improve discoverability and time-to-market for new products.
Quick Read
materials-science
6.1
Apr 23, 2026
Load-Dependent Hardness Models Improve Materials Screening
Researchers present a machine learning approach that predicts Vickers hardness as a function of indentation load using a large, curated set of experimental measurements. Models trained directly on experimental, load-dependent Hv outperform approaches that rely on DFT-derived proxies such as bulk and shear moduli, and multi-task models that mix experimental and computed targets. The work demonstrates that explicitly including the indentation load plus compositional, electronic, and structural descriptors is both necessary and sufficient to achieve accurate hardness predictions. The paper highlights the importance of high-quality experimental data and measurement metadata for reliable materials informatics and suggests practical changes for high-throughput screening workflows that currently use DFT-only proxies.
Quick Read
moire-superlattices
6.1
Apr 23, 2026
Machine Learning Reveals Ultralow-Barrier Sliding in Ferroelectric Moiré Superlattices
Using machine-learning-accelerated molecular dynamics, researchers show spontaneous thermally driven interlayer sliding in ferroelectric MoS2 moiré superlattices at room temperature, with relative velocities near 1 m/s. The motion is not rigid layer translation but a global drift of the moiré pattern enabled by a domain-wall-mediated collective reconstruction pathway that has an almost vanishing energy barrier. Allowing full structural relaxation reproduces the barrierless sliding, while sparse sulfur vacancies, at about 0.1%, trigger a transition from long-range sliding to localized pinning. The effect holds across twist angles and multidomain structures. Findings change how we should model switching and mechanical response in sliding ferroelectrics and highlight defect sensitivity for device stability and design.
Quick Read

Showing 24 articles

See the full news feed