Jiseki Health, Palo Alto, California
November 2017 – Present
- Developed algorithms, workflows and care plans for chatbot-driven clinical services (Flu, UTI, Mental health, etc.), improving access to basic clinical care for over 300,000 low-income users across the US.
- Created the Jiseki Visual Care Plan™, a new approach to physician instruction delivery that improves patient compliance and clinical outcomes. This IP is to be licensed to health systems.
- Led nationwide Telehealth strategy and policy research to identify business opportunities and regulatory barriers
- Facilitated physician-engineer communication and data-driven decision making through analytics and visualizations
Insight Data Science, San Francisco, California
September 2017 – December 2017
Health Data Science Fellow
- Combined patient data from multiple sources (imaging, genomic, clinical) to build Breast Cancer staging model that helps health plans identify and care for high-risk cancer patients faster and easier.
- Combined Tensorflow 3D Convolutional Neural Net (>120,000 breast MRI slices) with high-dimensional NGS data (RNASeq read counts with >23,000 genes) and clinical data to predict breast cancer stage with 78% Recall.
- Deployed models (written in Python) on Flask web app hosted on Amazon Web Services (AWS) EC2 for user access.
- Created SQL database using Postgres to store risk scores and query user data
- Coached non-healthcare teammates (engineers) on clinical concepts like ICDs, HL7, SNPs, CNVs, EKG and lab results.
CEP/MedAmerica, Emeryville, California
May 2017– August 2017
Data Science Summer Intern
- Discovered manual/time-consuming Readmission prediction process and pitched automated ML-driven alternative.
- Built 30-day Readmission Risk Prediction models using Gradient Boosting Machines (AUC 0.86) and Deep Learning (AUC 0.83) using the h2o platform in R that outperformed industry standard (0.68 AUC) by 18%.
- Leveraged in-depth understanding of clinical features to engineer >150 features in SQL (SSMS) using entire patient visit history, proprietary comorbidity flags and comorbidity scoring from over 23 million EMR and claims data.
- Upgraded proprietary STATA scripts for deriving CMS quality metrics (QCDR) to production-level R code for Inpatient and Emergency Department practice lines using over 62 million records of CMS HCUPS and client data.
University of San Francisco, San Francisco, California
November 2015–April 2017
- Analyzed over 4TB of RNASeq data using several Bioconductor packages in R to discover groups of differentially expressed cytokines with potential for drug development in ER-positive vs ER-negative breast cancer.
- Predicted the risk of Hepatocellular carcinoma in Python using cellular markers and alcohol consumption (91% Accuracy). Random Forest feature importance showed GGT and alcohol consumption were among top predictors.
- Evaluated n-grams of surveys responses in Python and R using NLP and queried PostgreSQL Medicare database to gain insight into patient pain points. The top 3 areas of concern were cost, service delays and mismanagement.
One Medix Health, Ibadan, Nigeria
September 2012 – April 2016
- Consulted with leading Health Systems and Medical schools to develop telehealth programs, learning management systems, increase provider efficiency, improve navigation, block revenue leaks and improve patient experience.
- Designed new registration and payment systems that cut queues and patient wait-times by 68%.
- Grew the team from 3 to 13 people and scaled customer-base nationwide increasing revenue by over 430%
University College Hospital, Ibadan, Nigeria
March 2012 – August 2014
- Rotated through Neurology, Surgical Oncology, Endocrinology, Hematology, Obstetrics, and Pediatric Oncology, getting a broader sense of the care continuum and compelling opportunities to improve quality of care through technology.
- Initiated 6 landmark projects to increase quality of care and clinical education including the patient Navigation map, Tele-surgery for medical education, the Clinical Simulation Center, and the Medical Student Mobile Companion App.