Artificial intelligence (AI) is revolutionizing healthcare, providing innovative tools for early disease detection and precise tracking of treatment outcomes. A recent study led by Yale researchers used AI technology to assess skin involvement and treatment response in patients with systemic sclerosis.
New AI Technology Tackles Systemic Sclerosis Diagnosis
Systemic sclerosis (SSc), also known as scleroderma, is a chronic autoimmune condition where the body produces excessive collagen, causing the skin and other tissues to harden. This thickening significantly affects patients’ quality of life. The disease can also impact internal organs, causing additional stress for patients.
“Systemic sclerosis can affect both the skin and internal organs, making the disease highly visible,” said Monique Hinchcliff, MD, MS, an associate professor of medicine at Yale and the primary investigator of the study. “Earlier diagnosis can allow for lifestyle changes and treatments to be initiated before internal organ damage occurs, leading to longer and healthier lives.”
Challenges in Traditional Diagnosis Methods
The current standard for assessing skin thickness in SSc patients is the modified Rodnan skin score (mRSS). While it is widely used, the mRSS has limitations, including its reliance on a pinch test, long intervals to detect meaningful changes, and its susceptibility to being influenced by obesity or edema.
Ilayda Gunes, the lead author of the study and a research assistant in Dr. Hinchcliff’s lab, emphasized that their goal was not to replace the mRSS but to find complementary methods that could provide more accurate and faster results for clinical trials. “We wanted methods that are quantitative and reproducible, which could shorten the duration of trials,” she said.
AI-Driven Fibrosis Score: A Breakthrough in Skin Biopsy Analysis
In their study, the researchers used deep neural networks (DNN), a form of AI, to analyze skin biopsies from SSc patients. The team generated a “fibrosis score” for each sample and became the first to apply AI to SSc skin biopsies.
The study compared the DNN-derived fibrosis score with the traditional mRSS in an SSc clinical trial. The results showed a weak correlation between the two, indicating that the DNN was capturing skin features beyond what the mRSS could detect through a pinch test. Additionally, the DNN detected different histologic features compared to the mRSS.
“The low correlation between the mRSS and fibrosis scores suggests that AI may be capturing aspects of the skin that clinicians cannot detect with a simple pinch test,” Gunes said.
Combining AI with Traditional Methods for Better Results
Given that the mRSS and fibrosis scores measure distinct pathological features, the researchers suggest that combining both methods may be more effective than using either one alone. This dual approach could help streamline clinical trials, accelerate recruitment, and improve participant diversity, ultimately improving the generalizability of SSc trial results.
Looking Ahead: AI’s Role in Advancing SSc Diagnosis
Dr. Hinchcliff believes that AI will play a major role in the future of SSc diagnosis. “AI is evolving rapidly, and we are testing new methods that could help measure the three components of SSc skin disease: inflammation, vascular abnormalities, and fibrosis,” she said.
The hope is that AI models can be trained to detect early signs of SSc through skin biopsies or chest CT scans, allowing for earlier intervention and preventing organ damage.
Related topics
