AWS Public Sector Blog

How startups PointHealth AI and Protege transform health treatment recommendations with AI on AWS

How startups PointHealth AI and Protege transform health treatment recommendations with AI on AWS

In today’s healthcare system, doctors often face a difficult question: how can they make the best treatment decisions without the full picture? Key datapoints—like patterns in treatment histories and outcomes across patient demographics—are often difficult to procure. Without a way to obtain this information, doctors risk recommending less effective treatments, leading to repeat visits, higher healthcare costs, and missed opportunities to improve patient care.

This challenge grows more pressing as healthcare needs grow. By 2030, approximately 67 million Americans will be enrolled in Medicare at an annual cost of $1.2 trillion, according to forecasts published in the Forum for Health Economics and Policy—putting unprecedented pressure on the healthcare system to get treatment decisions right the first time.

PointHealth AI, a healthcare technology (HealthTech) startup, and Protege, a data innovator specializing in artificial intelligence (AI)-ready datasets, are working together to tackle this issue. Using Amazon Web Services (AWS), their AI-powered solution combines de-identified demographic and clinical information from millions of patient records, delivering more precise and informed treatment recommendations. This data-informed strategy aims to give doctors the information they need to prescribe the right care for a patient’s unique needs—on the first try.

Meeting the need for AI-ready healthcare data

When PointHealth’s founders set out to build their AI-driven treatment recommendation system, they quickly hit a roadblock. Traditional data providers typically focus on medical device testing or US Food and Drug Administration (FDA) regulatory studies rather than providing training data for AI models. “We went to many different data providers, and most of the companies we talked to were not super interested in us creating models from healthcare data,” explains Joe Waggoner, Chief Executive Officer (CEO) and co-founder of PointHealth. “That’s the great thing about Protege—their whole mission is to provide training sets for companies building products with AI in mind. Without that data, I couldn’t get the company off the ground.”

Transforming raw data into actionable healthcare insights

Protege’s system transforms raw healthcare data into AI-ready training sets while maintaining strict compliance with the Health Insurance Portability and Accountability Act (HIPAA). Through its healthcare division, Protege Health, Protege works with close to 20 data partners to collect healthcare information from multiple sources. Its solution processes an unprecedented volume of data: 30 billion electronic medical records, 40 million medical images, and claims data covering almost the entire United States, with 90% historical coverage of mortality data.

The process starts with helping PointHealth discover and procure the right set of data, followed by custom transformations based on specific AI use cases. This transformed data is what PointHealth uses to train its reinforcement learning models to analyze patient similarities and recommend optimal treatment methods for patient needs.

“We use genomics, drug interactions, and patient similarity to come up with a course of treatment that is most likely optimal for a specific patient,” says Rachel Gollub, Chief Technology Officer (CTO) and co-founder of PointHealth. “When a doctor enters patient information from their electronic health record system, our model can compare their data against millions of similar cases to recommend the most effective treatments.”

Building a comprehensive data solution with AWS

To process and store this unprecedented volume of healthcare data securely, Protege built their infrastructure on AWS. The solution incorporates multiple AWS services, including Amazon Elastic Compute Cloud (Amazon EC2) for a variety of tasks, Amazon EMR for running Apache Spark jobs, Amazon Simple Storage Service (Amazon S3) for secure storage, AWS DataSync for seamless data transfer, and Amazon Virtual Private Cloud (Amazon VPC) and AWS Network Firewall for security.

“We’ve had a very positive experience with AWS,” says Richard Ho, CTO at Protege. “The entire support team has been great, especially in helping us develop a robust solution for data delivery that works for both us and PointHealth.”

AWS enables rapid scalability and compliance with healthcare industry regulations. For PointHealth, this means they can securely process large datasets without managing complex compliance requirements. “The ability to auto-scale in a HIPAA-compliant way is invaluable,” explains Gollub. “We started with a larger initial database than most companies ever get to work with, and AWS helped us accommodate that scale seamlessly.”

Supporting healthcare providers and improving patient outcomes

PointHealth is preparing to launch their solution with smaller healthcare agencies, initially focusing on behavioral health systems with 20-30 clinics in a region. Their platform will provide doctors with evidence-based treatment recommendations for patients matching de-identified demographic and clinical data—while giving doctors final say in treatment selection.

“We want to allow doctors to operate at the top of their license,” says Waggoner. “There’s no way a human can remember millions of cases and all the patient similarities. So, we’re taking that burden and offering this technology to doctors to help them make better treatment decisions.”

By integrating seamlessly with existing electronic health record systems, PointHealth makes it simple for healthcare providers to incorporate AI-driven insights into their current workflows.

The future of AI in healthcare

Both companies see significant potential for AI to transform the healthcare sector. They’re exploring multimodal approaches, combining reinforcement learning with generative AI to enhance treatment recommendations and automate administrative tasks like medical coding.

“Doctors are often burdened by desktop medicine and don’t have enough time to see patients,” says Ho. “There’s so much that technology can do to give doctors the right tools to provide the best treatment paths, which ultimately flows down to better patient outcomes.”

Accelerating healthcare innovation with AWS

The collaboration between Protege and PointHealth demonstrates how AWS can support leading-edge healthcare solutions that benefit both providers and patients. Using cloud technology and AI, these companies are working to help patients receive the most effective treatments from the start, potentially reducing repeated visits, and improving overall healthcare efficiency.

Healthcare professionals and technology leaders can learn more about this innovative approach to treatment recommendations by visiting PointHealth and Protege at their shared booth at the upcoming Precision Medicine World Conference.

Learn more about AWS for HealthTechs

AWS continues to support healthcare innovation across the industry, from generating new therapeutic candidates, to matching patients with clinical trials and powering patient engagement applications. The AWS for HealthTechs hub provides the services, data, models, and secure infrastructure needed to scale generative AI across healthcare organizations.

Discover how generative AI can accelerate health innovations and improve patient experiences by visiting the AWS for HealthTechs hub.

Read related stories on the public sector blog: