Automated Interpretation of Medical Imaging
Our initial goal is to develop artificial intelligence applications that will facilitate and in some cases even automate medical imaging diagnosis (e.g. for radiology, histopathology or ophthalmology). Recent advances in computing infrastructure as well as AI algorithms have allowed for tremendous progress in automating the interpretation of medical diagnostic imaging.
Prediction of Medical Risk
Projects also include building predictive algorithms of patients’ medical risk, leveraging medical record and billing claim data. This will allow for prospective identification of patients in higher need of preventive and/or therapeutic interventions. This initiative can support primary prevention, but also earlier diagnosis of disease, or prevention of disease complications.
Improvement of chronic disease care using real world data
Generalizability of clinical trial results has been acknowledged to be a challenge, since the characteristics of the selected research subject population can bias the assessment of clinical outcomes for other patient populations. Creating the ability to predict results in target populations or individuals who were not necessarily represented in the clinical trial, will add significant value.
Prediction of risk for medication non-adherence
Becoming non-adherent to long term medications can adversely affect the prognosis for chronic diseases, such as diabetes or heart failure. Historically, medication non-adherence has been identified only retrospectively, oftentimes too late to prevent worsening of these medical conditions. However, predicting one’s future risk of sub-optimal adherence or even discontinuation of medications is more achievable today, due to advances in our knowledge, data sources and technology.
AI-informed diagnostic support in Intensive Care Units
Currently, the practice of ICU medicine relies mostly on the ability and experience of clinicians to integrate, assess and make decisions based on various types of information. To complicate this task even further, the data changes in real time and clinical decisions need to be amended accordingly. Due to the complex data sources, immediate decision needs as well as vulnerability of these patients, ICU is the environment where AI-coordinated diagnostics would potentially add great value.