AI shows promise in predicting the progression of autoimmune disease
Autoimmune diseases, in which the immune system mistakenly attacks the body’s own healthy cells and tissues, often include a preclinical stage before diagnosis that’s characterized by mild symptoms or certain antibodies in the blood. However, in some people, these symptoms may resolve before culminating in the full disease stage.
Knowing who may progress along the disease pathway is critical for early diagnosis and intervention, improved treatment and better disease management, according to a team led by researchers from the Penn State College of Medicine that has developed a new method to predict the progression of autoimmune disease among those with preclinical symptoms.
The team used artificial intelligence (AI) to analyze data from electronic health records and large genetic studies of people with autoimmune disease to come up with a risk prediction score. When compared to existing models, this methodology was between 25% and 1,000% more accurate in determining whose symptoms would move to advanced disease.
The research team has published their findings in the journal Nature Communications.
“By targeting a more relevant population—people with family history or who are experiencing early symptoms—we can use machine learning to identify patients with the highest risk for disease and then identify suitable therapeutics that may be able to slow down the progression of the disease. It’s a lot more meaningful and actionable information,” said Dajiang Liu, distinguished professor, vice chair for research and director of artificial intelligence and biomedical informatics at the Penn State College of Medicine and co-lead author of the study.
Approximately 8% of Americans live with autoimmune disease, according to the National Institutes of Health, and the vast majority are women. The earlier you can detect the disease and intervene, the better, Liu said, because once autoimmune diseases progress, the damage can be irreversible. There are often signs of the disease before an individual receives a diagnosis. For example, in patients with rheumatoid arthritis, antibodies can be detected in the blood five years before symptoms begin, the researchers explained.
The challenge with forecasting disease progression is sample size. The population of individuals who have a specific autoimmune disease is relatively small. With less data available, it’s harder to develop an accurate model and algorithm, Liu said.
To improve prediction accuracy, the research team developed a new method, dubbed Genetic Progression Score or GPS, to foretell the progression from preclinical to disease stages. GPS leverages the idea behind transfer learning—a machine learning technique where a model is trained on one task or dataset and then fine-tuned for a different but related task or dataset, explained Bibo Jiang, assistant professor of public health sciences at the Penn State College of Medicine and lead author of the study. It allows researchers to glean better information from smaller data samples.
For example, in medical imaging, artificial intelligence models can be trained to tell whether a tumor is cancerous or non-cancerous. To create the training dataset, medical experts need to label images one-by-one, which can be time-intensive and limited by the number of images available.
Liu explained that instead, transfer learning uses more numerous, easier-to-label images, such as cats and dogs, and creates a much larger dataset. The task can also be outsourced. The model learns to differentiate between the animals, and then it can be refined to distinguish between malignant and benign tumors.
