top of page


We are seasoned data scientists with a long track record of delivering results in healthcare and tech.

Coming Soon: Learning Engine

A personalized syllabus for you to learn any skill of your choosing. A course on French cooking technique? A course on basic car maintenance? What about a course on Charles Darwin’s life? All of these are possible with our education engine. Put the power of AI to work for you and learn from all the internet has to offer with a curated syllabus on any topic. Input what you want to learn and we will provide a course of articles and videos to master any topic you want.

Medication Adherence Prediction and Intervention

Medication adherence is a huge problem for patients with chronic illnesses. Our team built a custom solution for CVS Health to predict which patients are at risk of non-adherence to their medications and proactively intervene in their healthcare journey. Through the use of ensemble learning, experimentation, and multi-armed bandits we were able to drive tens of millions of dollars in savings via reduction in preventable illness progression.


Personalized Location Recommendation for Medical Intervention

Preventive interventions are key factors in reducing patients risks and avoiding events such as flue outbreaks, inpatient hospitalization or self harm events such as suicidal attempts. One key point in preventive interventions is wether or not patients or targeted population will change their behavior when outreached through a low touch point channels such as emails, sms, mail & etc. 

Through scalable machine learning and unsupervised learning, our team build a location recommendation engine that would identify patients travel behavior using home/work address and care information and recommend a personalized locations that the patient is most likely to receive/willing to receive care in. Using this engine as a basis for preventive intervention we were able to increase the intervention behavior change rate and  by 5% and save CVS Health more than $7 Million in one year.  

Colorectal Cancer Screening Model & Program

When caught early, colorectal cancer is both highly survivable and inexpensive to treat; when not caught in time, survival rate can drop as low as 5% and the cost to treat can increase by two orders of magnitude. While colonoscopy is the gold standard for early detection and polyp removal, the invasiveness of the procedure, the fear of associated pain, and logistical challenges make some patients reluctant to undergo a preventive screening, thereby exposing them to elevated risk of late detection.


We built a predictive model that used embedding-generated features and gradient boosting to understand, among patients who were not up to date with US Preventive Services Task Force recommendations on colonoscopy, which ones would be good candidates for a non-invasive fecal immunochemical test (a "home kit") as an alternative screening method. Offering patients who were extremely reluctant to undergo a colonoscopy an alternative way to screen reduced the population's (n = 120k) rate of getting late stage colorectal cancer by almost 30%, and saved CVS over $6M per year in treatment costs.

Depression Detection and Suicide Prevention

Using the power of advance Machine Learning and Statistical Analysis our team in collaboration with CSV Health was able to identify individuals who are at high risk of depression and suicidal ideation in advance (up to three months) and  able to make preventive measures such as therapy and medication to avoid suicidal attempt and hospitalization. We used Electronic Medical Records and insurance data to identify those high risk individuals and try to get them to adhere to their medication and attend therapy to make sure they get the help they need. This effort was able to save 132 people in span of one year who were ideating suicidal though and some who had history of suicidal attempts.

bottom of page