AI Toe Scan Quickly Detects Hidden Artery Disease

Peripheral artery disease, or PAD, is a widespread condition that often goes unnoticed, yet it impacts blood flow throughout the body.

This disease develops when fatty deposits, known as plaque, accumulate inside arteries, causing them to narrow over time. As a result, less blood reaches the legs and feet.

Approximately 8 to 12 million Americans have PAD, with millions more worldwide affected. Despite its prevalence, many people remain undiagnosed until the disease advances to a more serious stage.

Some individuals experience pain during walking or exercise, others may notice numbness, weakness in their legs, slow-healing wounds, or skin discoloration. Many individuals, however, show no noticeable symptoms at all.

This delay in diagnosis can be risky. PAD is a leading cause of limb amputations because insufficient blood flow can damage tissue and hinder wound healing. It’s also strongly associated with heart disease and stroke, making early detection critical for public health.

Today, doctors typically diagnose PAD using the ankle-brachial index (ABI) test, which compares blood pressure readings from the ankles and arms to identify circulation issues. While reliable, this process can be slow and requires special equipment and trained personnel.

Researchers at the University of California, San Diego, believe they’ve discovered a quicker, simpler screening method. Their new research indicates that using a technology called photoplethysmography (PPG), coupled with artificial intelligence, can accurately identify signs of PAD.

The study was published in npj Digital Medicine.

Photoplethysmography, or PPG, is already used in many health devices today. It works by shining light onto the skin and measuring how much light bounces back. Since blood flow influences how light behaves inside tissues, analyzing these signals can reveal circulation details.

Most people have used PPG without realizing it. Pulse oximeters clipped onto fingers in hospitals utilize this technology to measure oxygen levels and pulse rate.

For this study, researchers placed a light sensor on patients’ toes to capture blood flow signals. They then employed artificial intelligence to analyze patterns within these signals, looking for indicators of PAD.

The project started when Dr. Mattheus Ramsis, a cardiology informatics expert at UC San Diego, learned that toe PPG signals were already being recorded during routine ABI tests. He saw potential in these signals as standalone diagnostic markers.

The team collected over 10,000 toe PPG recordings from more than 3,500 patients treated at UC San Diego Health between 2020 and 2025. Using this large dataset, they identified numerous features within the signals linked to blood flow abnormalities.

They then developed a machine learning model capable of predicting PAD presence solely based on the PPG data.

The AI system performed impressively, correctly identifying PAD cases about 83% of the time. In comparison, traditional risk assessments typically reach only 60% to 65% accuracy.

The researchers found that including patients’ smoking history slightly improved the model’s accuracy. Since smoking significantly damages blood vessels and reduces circulation, it’s a critical risk factor for PAD.

Importantly, the model performed consistently across diverse patient groups, including different races—Black, Hispanic, and White patients—as well as individuals with diabetes, coronary artery disease, or kidney failure.

The team envisions this method making PAD screening more accessible. Since PPG tech is integrated into smartphones, wearables, and blood pressure monitors, future iterations could enable individuals to screen themselves at home.

This approach could be especially beneficial for underserved populations who face barriers like transportation, cost, or limited access to healthcare. Earlier detection might allow for prompt treatment, helping prevent severe outcomes like amputations.

Dr. Ramsis clarified that this isn’t meant to replace ABI testing but act as a quick, preliminary screening tool to identify those needing further evaluation.

This research aligns with a growing focus on digital diagnostics in medicine. AI analysis of signals related to heartbeat, respiration, or blood flow could facilitate earlier, cheaper disease detection in the future.

Overall, the study is promising, leveraging existing technology that is already widespread. The sizable patient data and consistent results across racial groups build confidence in the findings.

Nonetheless, since the study was conducted within a single healthcare system, further testing in broader populations and with different devices is necessary. The accuracy, though impressive, remains below comprehensive diagnostic methods, indicating that this system is best suited for initial screening rather than definitive diagnosis.

Despite these limitations, the development of AI-based blood flow screening has the potential to become an important tool in preventing serious complications related to PAD.

If you’re concerned with cardiovascular health, consider exploring studies on how drinking milk may influence heart disease risk, or how herbal supplements might affect heart rhythms.

For additional insights, see recent research on how espresso impacts cholesterol levels, and findings suggesting Vitamin K2 might help lower heart disease risk.

Source: University of California, San Diego.