Clinical data science
Develop predictive models from longitudinal records and structured clinical data, with endpoints and evaluation plans aligned to the research question.
MLAIA Medical AI
Clinical context.
Scientific rigor.
AI for clinical records, medical images and biosignals. We help hospitals and MedTech teams turn research questions into models that can be evaluated, understood and integrated.
A specialist practice of MLAIA Data ScienceLed by Dr. Yochai Edlitz · Ph.D., Weizmann Institute
Where we help
Depth where it matters.
Focused expertise, with the research and engineering support to move your project forward.
Medical imaging
Work with image and spectral data, including segmentation, feature extraction and model evaluation across relevant cohorts.
Biosignal analysis
Apply signal processing and machine learning to physiological measurements, with attention to measurement quality and repeatability.
Clinical NLP & LLMs
Explore information extraction and clinical-text workflows. Define grounding, review and evaluation requirements around the intended use.
Validation, bias & causal analysis
Assess leakage, cohort effects, confounding and generalization. Report uncertainty and limitations alongside model performance. For questions about intervention effects, assess causal study design and assumptions. Explore our causal-analysis approach.
Hospital data infrastructure
Support research environments and clinical-data pipelines, including EMR and FHIR integration requirements and secure deployment planning.
Client perspective
Experience in hospital innovation.
“Your leadership, strategic thinking, and excellent communication skills were instrumental in successfully executing this project.”
Dr. Gil Levy — Head of Innovation, Assuta Ashdod Medical Center
Excerpt from the testimonial published on MLAIA’s website.
Medical & sensitive data
Prepare useful data.
Control sensitive information.
De-identification workflows for structured records and free text, designed around the intended use and access model. Combine identifier detection, masking or pseudonymization, evaluation and review of residual disclosure risk.
Define what to protect
Identify direct identifiers and contextual information that could expose an individual. Establish rules for the data types, language and downstream task.
Evaluate the transformation
Measure missed identifiers and unnecessary removal on representative examples. Assess how the changes affect the usefulness of the data.
Connect to approved AI workflows
Define what information may reach an LLM or multimodal service, where it may be processed and how access is controlled. De-identification alone is not a guarantee of anonymity.
How we work
A clear path from question to evidence.
Frame the problem
Agree on the decision, available data, operating constraints and success criteria.
Establish a baseline
Inspect data quality, design the evaluation and test what current approaches can do.
Develop & evaluate
Build in bounded milestones. Review results, failure modes and limitations together.
Integrate & hand over
Plan deployment, monitoring, documentation and ownership around the agreed scope.
From our knowledge base
Explore the thinking behind the work.
Before we start
Good questions. Straight answers.
Do you provide a finished, approved medical device?
This practice provides AI research, engineering and integration services. Regulatory status and clinical-use authorization are specific to the product and intended use; they are not implied by engaging MLAIA.
Can you work with small or multi-site datasets?
We can assess feasibility, data quality, cohort composition and evaluation design. Small or heterogeneous datasets may limit reliable conclusions; those limitations need to be made explicit.
How is sensitive clinical data handled?
Data access, hosting, de-identification, security controls and contractual requirements are agreed with the data owner before work starts. Please do not submit patient information through the website contact form.
Medical AI
Tell us what
you’re working on.
Share the problem, the data you have and what success would look like. We’ll discuss a practical next step.
Please don’t include confidential datasets, credentials or patient information.