Peripheral Artery Disease, or PAD, affects as many as 6.5 million people in the United States. Patients with PAD are at increased risk of coronary artery disease and cerebrovascular disease like strokes — and in extreme cases can lead to limb amputation. The Stanford Data Science team has developed an AI model that identifies suitable primary care patients to refer to cardiovascular health specialists for further evaluation.
According to the project’s charter, the model, which went live May 15, is capable of identifying patients who are likely to have undiagnosed PAD, using patient data such as diagnosis, procedures, labs, and prescriptions.
“Primary care settings are central to the early detection of peripheral artery disease, yet its nonspecific symptoms and variable presentation often hinder timely diagnosis,” noted Eri Fukaya, MD, a Vascular Medicine specialist at SHC and a Clinical Professor of Surgery and Medicine at Stanford University School of Medicine who helped develop the model. “Proactive screening, education, and early intervention are key to mitigating serious complications. Now is a crucial opportunity to leverage artificial intelligence to enhance early identification and improve patient outcomes.”
Within primary care, “PAD is often overlooked,” added Stanford Health Care family medicine physician Andrew Schechtman, MD. “A tool that proactively identifies high-risk, symptomatic patients and presents them to busy primary care clinicians along with an easy, few-click path to comprehensive evaluation and treatment has the potential to transform care."
Primary care settings are central to the early detection of peripheral artery disease, yet its nonspecific symptoms and variable presentation often hinder timely diagnosis."
- Eri Fukaya, MD, Clinical Professor of Surgery and Medicine at The Stanford University School of Medicine
“The information is highly actionable. Based on recently published American College of Cardiology/American Heart Association guidelines, there are a number of high impact steps to take to improves the lives of patients with PAD,” said cardiologist Sneha S. Jain, MD, MBA. “However, current risk assessment tools are too broad — we needed a way to efficiently identify patients with undiagnosed, symptomatic PAD in SHC’s Primary Care population in order to provide personalized and guideline-directed management.”
Abby Pandya, Senior Manager for Data Science Product Management, TDS, goes into more detail about their team’s groundbreaking work on SHC’s Peripheral Artery Disease Model. “A lot of us are familiar with coronary artery disease, in which the vessels in our hearts are getting clogged up over time,” said Pandya.
“PAD is similar, but it’s in the extremities, such as the legs. Peripheral artery disease is often undiagnosed and under-treated. Awareness of it is low, even in clinical settings and among physician populations and patient populations.
“So the whole goal here is to evaluate and expand this AI-guided workflow forward,” Pandya continued. “We have a model running on Stanford Medicine’s population and we are now using that to drive a patient questionnaire to screen for symptoms and help providers in identifying the patients.”
Since patients may also face comorbidities like diabetes and hypertension, there may be time to start them on a different track such as behavior modification, exercise, or medication, after being diagnosed via an ankle brachial index/toe brachial index (ABI/TBI) diagnostic evaluation.
“An ABI or TBI takes the ratio of blood flow in your extremities and compares it to your normal arm blood pressure and blood flow. That ratio determines whether you have PAD, how severe it is, and what kind of course of action would be beneficial,” explained Pandya.
“The whole goal is to try to identify patients who may have the disease who may benefit from this noninvasive screening — and then, if it’s clinically necessary and makes sense, to start getting them on a course of action to prevent bad outcomes,” she said.
The achievement is a result of the dedicated efforts of our clinical, operational, and technical teams. The Stanford Data Science team built on the work of Dr. Elsie Ross in collaboration with vascular specialists Dr. Eri Fukaya and Ani Bagdasarian and Stanford Medicine Partners Primary Care Dr. Andrew Schechtman and Hilary Garrigan to release an AI-augmented workflow that identifies suitable primary care patients to refer to vascular medicine specialists:
- Model provides prediction using patient data
- Ask patient if symptomatic via MyHealth or during rooming
- Notification to PCP, if symptomatic
- Refer to Vascular Clinic, if appropriate
- Patient scheduled for ABI and subsequent follow-up with Vascular Medicine
This project demonstrates our ability to build, deploy, and monitor an in-house AI solution using SHC’s Responsible AI Life Cycle. It enhances our capacity to identify and manage patients with PAD through an AI-Augmented workflow and includes a monitoring plan to evaluate system integrity, model performance, and operational impact.
