From Procedural Medicine to
Privacy-Preserving Personalized Medicine

Dr. Ruffin has been writing about digital health, precision medicine, and personalized health for decades. He helps clients learn to acquire information and clinical technologies so that they can perform procedures on individual patients as effectively and efficiently as possible AND use the data collected to improve predictions of clinical and financial outcomes for all patients. Dr. Ruffin helps clients identify and select use data from their information systems to build platforms for privacy-preserving personalized care with other like-minded health care organizations.

Practicing Privacy-Preserving Personalized Medicine

About 5% of women delivering babies in hospital suffer post-partum hemorrhage of more than 1 liter of blood. The best predictive models for post-partum hemorrhage in the published literature have an accuracy (area under (a ROC) curve) of less than 0.6. Imagine that your organization could predict, with an accuracy (area under (a ROC) curve) of better than 0.85, which women will suffer post-partum hemorrhage when they are first admitted to Labor and Delivery. Dr. Ruffin helped to organize data scientists at a major medical center, collaborating with obstetricians in three major health systems, to build privacy-preserving predictive models of post-partum hemorrhage with this high degree of accuracy, allowing physicians and nurses to focus their efforts on high-risk patients to prevent this complication of pregnancy. This is what we mean by privacy-preserving personalized medicine.

Privacy-preserving personalized medicine requires accessing all data in patients’ records, including identities of health care facilities, clinicians treating them and dates and times of service, data usually scrubbed from “de-identified” data sets, in a way that does not jeopardize patients’ privacy or conflict with HIPAA. Dr. Ruffin advocates the use of “privacy-preserving” architectures for predictive analytics such as those described by Andrew Trask and the OpenMined not-for-profit community (www.openmined.org) he helped to create. Three technologies, unfamiliar to most health care systems, are necessary for privacy-preserving personalized medicine, including federated learning, differential privacy and homomorphic encryption. The purpose of using information systems in health care is greater than registration, scheduling, billing, and retrieval of laboratory and radiology results. Our purpose is to predict accurately the future for patients, based on their demographics, physiology, ailments, treatments, and genetics using privacy-preserving personalized medicine.

OpenMined is an open-source community whose goal is to make the world more privacy-preserving by lowering the barrier-to-entry to private AI technologies.

We are current with cutting-edge technologies and can help to organize the right people to create the software functions and technical architectures privacy-preserving personalized health care. The purpose of using information systems in health care is greater than registration, scheduling, billing, and retrieval of laboratory and radiology results. Our purpose is to predict accurately the future for patients, based on their demographics, physiology, ailments, treatments, and genetics.

Doctor Discussing With Patient