GLIF3 - Computer Interpretable Guideline (CIG) Language
During 1999-2003, While at Stanford, I was part of the InterMed Collaboratory (Stanford, Harvard, Columbia) which developed the GLIF3 computer-interpretable guideline (CIG) modeling language, and enactment engine. I implemented a diabetic foot guideline using GLIF3 at RAMBAM hospital.
During 2011-2015 I led a large European project with 60 participants from 13 organizations (Acedemia, industry, hopsitals) in 5 European countries. We developed a personalized and distributed decision-support system with mobile sensors for patients and their care providers. Mobiguide improved health care for atrial fibrillation and gestational diabetes patients.
In this ISF project (2016-9) we are developing a framework for combining Computer-interpretable guidelines with OWL medical ontologies to plan therapy for patients with multimorbidity. See our AMIA paper. This project is now continuing as part of CAPABLE (see below).
CAncer PAtients Better Life Experience
This new EU Horizon 2020 project 2020-2024 extends the knowledge-based decision support with data-based machine learning models to learn what is best to improve the quality of life of chronic patients. Topics explored include semantic data integration, machine learning to improve care processes, GoCom, and Patient Coaching System that uses psycho-behavioral theories to increase patient engagement in their health and improve their quality of life. Project web site סרטון בעברית
Ontology-guided mining for prognosis
Children with developmental disorders often have multiple diagnoses, which progress over time. Aided by a literature-based ontology, we used Self-Organizing Maps to detect clusters of children with particular multiple diagnoses and then applied Sequence Mining to find their progression.
Patient-centric Data Science
Our group's mission is to use data science methods in order to improve people's health by developing methodologies and tools that provide decision-support to patients and their care providers. To do so, we combine methods for knowledge representation and reasoning, data management, and statistical inference and reasoning and apply it to large longitudinal multidimensional health data sets that are continuously updated. more
Learning Context & Process
Clinical guidelines may recommend the same treatment for different patients. But some subpopulations may not benefit from the first-line treatment (e.g., antibiotic A) but from a different treatment (Anti-c). Can we use machine learning to cluster similar patients belonging to different clinical context? These can be different wellbeing states (good vs. poor wellbeing). Can we predict the most effective care process for patients belonging to a certain context?
"SHOMER MASSA": Safeguarding and promoting mental health and resilience among adolescents and young adults using mobile technology. Collaboration with Prof. Danny Koren from Psychology Dept. and graduate students Ofri Ben-Yehuda and Efrat Dreazen.