Distilling the best in healthcare resources-- so you don’t have to.Here is the website: https://evanbrociner.github.io

Helping bring together 200+ Raleigh-Durham-Chapel Hill healthtech leaders once a month for an in-person event to hang out.

I failed a lot. I learned a lot.

I led the development of an AI system to improve how healthcare providers handle prior authorizations. My approach included using OpenAI's GPT-3.5 for its language understanding capabilities, Microsoft Azure for managing data and computing needs, and an OCR model to process unstructured text. A key part of my role was explaining this technology to our customers and using their feedback to make the system better.Key Achievements:• Guided a cross-functional team consisting of engineers, product leaders, and healthcare experts to build an AI-driven prior authorization reviewer, employing technologies such as OpenAI GPT-3.5, Azure, and Hugging Face.• Implemented a robust engineering pipeline to structure text from over 20,000 unstructured PDFs, making them suitable for input into a LLM.• Spearheaded the recruitment efforts to build a top-notch team of machine learning engineers specializing in LLMs in healthcare.• Closely partnered with the product team to develop user-centric solutions for both healthcare providers and insurance companies, contributing to the overall product strategy and roadmap.

Consultant for the Mayo Clinic's Health Equity division, where I provided expertise and analysis to address disparities in healthcare access and outcomes.Key Achievements:• Conducted comprehensive analyses of health disparities at an enterprise level, identifying and quantifying disparities in healthcare access, treatment, and outcomes across diverse patient populations.• Utilized SQL to merge Electronic Health Record (EHR) data from over one million patients, enabling a detailed examination of Social Determinants of Health (SDOH) disparities among patients diagnosed with diabetes. This analysis provided valuable insights into the factors influencing health outcomes and guided targeted interventions to address disparities.• Developed a logistic regression model using R to assess odds ratios associated with various demographic and clinical factors among a large patient population of over 1.7 million individuals. This model provided a deeper understanding of the factors influencing healthcare outcomes and supported evidence-based decision-making in clinical practice and policy development.

I spearheaded AI-driven research on healthcare disparities, managed large datasets, and developed innovative diagnostic models, while also providing expertise in study design and statistical analysis.Key Achievements:• Led an article for JAMA Pediatrics studying how patient no-show rates differ based on income and whether visits are in-person or via telehealth at outpatient clinics. Recognized by the GI Division Leadership through an invitation to present study findings, showcasing expertise and insights.• Led an initiative resulting in the development of a highly accurate diagnostic AI computer vision model for pediatric appendicitis patients. Utilized PyTorch and over 100,000 ultrasound images as a way to significantly improve disease detection and diagnosis.• Provided leadership and expertise in study design and statistical analysis for over six research projects with clinical colleagues. Ensured thorough methodologies and accurate data interpretation.• Managed more than a dozen large patient datasets from Boston Children's Hospital, showcasing strong data management skills and effective handling of complex information.• Directed the development of a novel computer vision model framework for categorizing ultrasound images, utilizing Google Cloud's Optical Character Recognition (OCR). Established an efficient process for developing AI-driven ultrasound disease detection models, enhancing diagnostic accuracy and efficiency.

As a college student, I gained hands-on experience by shadowing physicians and contributing to clinical research efforts. This included utilizing the hospital's electronic health record (EHR) system to collect and organize data for studies on inflammatory bowel disease, aiding in treatment guideline development.Key Achievements:• Used Boston Children's Hospital's EHR system to collect and organize data for clinical studies on inflammatory bowel disease. Contributed to developing better treatment guidelines by analyzing patient information.• Led a project at Boston Children's Hospital analyzing lab results from the EHR to gauge the severity of inflammatory bowel disease. Helped the team make more informed treatment decisions for patients.

Under the guidance of Professor Zak Kohane, MD, PhD, I actively contributed to innovative population health studies aimed at improving patient health outcomes.Key Achievements:• Played a key role in developing experimental designs for population healthcare studies involving over 80 million patients. By contributing to the design process, I ensured scientific rigor and study validity, leading to reliable outcomes.• Designed and implemented robust R scripts to extract information from extensive patient datasets using SQL queries. These scripts efficiently retrieved and organized large-scale data, facilitating comprehensive analysis and supporting evidence-based decision-making.• Created R scripts to visualize complex data insights using ggplot2.• Communicated complex computer science experimental results effectively to diverse audiences. Facilitating their understanding and promoting data-driven decision-making in population health.

During my summer internship, I actively contributed to bioinformatics research led by a clinical Gastroenterology, Nutrition, and Hepatology fellow. As a result of my dedication and skills, I transitioned into a full-time employee after graduating as an undergraduate.Key Achievements:• Co-authored and presented three abstracts at the 2018 conference for the North American Society for Pediatric Gastroenterology, Hepatology & Nutrition, highlighting our research findings and contributions to the field.• Demonstrated proficiency in technical skills by creating and implementing Natural Language Processing (NLP) algorithms using custom Python scripts. These algorithms efficiently analyzed clinical notes from Boston Children's Hospital patients, enabling automated extraction of crucial information and valuable insights from unstructured data.• Played a pivotal role in developing the Boston Children’s Hospital Appendix Database, a significant contribution to improving care for appendicitis patients. Co-creating this database, which comprises over 4,000 cases, has empowered healthcare professionals with valuable insights and improved clinical decision-making.