Summary: Using artificial intelligence, researchers have created a Genetic Progression Score ( GPS) to track the progression of autoimmune diseases from preclinical symptoms to full disease. In contrast to previous models, the GPS unit integrates genetic information and electronic health records to provide personal risk values, increasing forecast correctness by 25 % to 1 %.
This technique allows for earlier diagnosis and better condition control, enabling earlier treatments. Providing a breakthrough in personal medicine and health equity, the framework could also be applied to investigate other underrepresented diseases.
Important Information:
- To improve the prediction of disease progression, GPS combines genomic studies and biobank data.
- The design outperformed 20 different methods, improving forecast accuracy considerably.
- Early detection of at-risk citizens makes for effective treatment and treatments.
Origin: Penn State
Autoimmune diseases typically have a preliminary step before treatment that is characterized by mild symptoms or specific antibodies in the bloodstream. This stage is where the immune system mistakenly attacks the body’s own good cells and tissues.
But, in some people, these signs may fix before culminating in the entire disease stage.
According to a group led by researchers from the Penn State College of Medicine, which has developed a novel method to predict the development of autoimmune disorder among those with experimental symptoms, knowing who does advance along the disease path is crucial for early diagnosis and treatment, improved therapy, and better disease management.
The team used artificial intelligence ( AI ) to compile a risk prediction score from data from large genetic studies of people with autoimmune disease and electronic health records to analyze it.
This methodology was between 25 % and 1, 000 % more accurate in determining whose symptoms would lead to advanced disease when compared to existing models.
The research team published their findings this week ( Jan. 2 ) in the journal , Nature Communications.
We can employ machine learning to identify individuals who are most susceptible to the disease and discover appropriate treatments that may be able to slow down the progression of the disease by targeting a more important people, such as those with family histories or those who are experiencing early signs.
” 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.
Before a person is diagnosed with the disease, there are frequently symptoms. For example, in patients with rheumatoid arthritis, antibodies can be detected in the blood five years before symptoms begin, the researchers explained.
The difficulty with forecasting disease progression is the sample size. A particular autoimmune disease has a relatively small population. With less data available, it’s harder to develop an accurate model and algorithm, Liu said.
The research team created a new technique known as the Genetic Progression Score ( GPS) to track the progression from preclinical to disease stages in order to improve prediction accuracy. Bibo Jiang, assistant professor of public health sciences at the Penn State College of Medicine and lead author of the study, explained the concept behind transfer learning, a machine learning method that allows a model to be trained on a single task or dataset before being fine-tuned for a different but related task or dataset.
It enables researchers to extract more valuable information from smaller sample sizes.
For example, in medical imaging, artificial intelligence models can be trained to tell whether a tumor is cancerous or non-cancerous. Medical experts must label images one-by-one in order to create the training dataset, which can take time and be hampered by the number of images available.
Liu said, 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 distinguish between the animals before being able to distinguish between benign and malignant tumors.
” You don’t need to train the model from scratch,” Liu said.
The” transferable” way in which the model separates elements from an image to determine whether it’s a cat or dog. You can modify the model to distinguish between a tumor and a piece of normal tissue with some adaptation.
In order to identify genetic differences between people with a particular autoimmune disease from those without it and to identify potential risk factors, GPS uses data from large case-control genome-wide association studies ( GWAS ), a widely used method in human genetics research.
It also incorporates data from electronic health record-based biobanks, which contain rich information about patients, including genetic variants, lab tests and clinical diagnoses.
This information can help identify individuals in preclinical stages and describe the progression stages from preclinical to the disease stage. The GPS model is then refined using data from both sources and incorporates elements that are pertinent to the actual spread of disease.
Large case-control studies and biobanks were able to draw advantages from the large sample sizes of case-control studies and improve prediction accuracy, Liu said, adding that those with high GPS scores are more likely to progress from preclinical to disease stages.
The team compared the GPS risk scores to those from the Vanderbilt University biobank, a health data initiative of the National Institutes of Health, using real-world data to forecast the progression of rheumatoid arthritis and lupus.
Compared to 20 other models that only use biobank or case-control samples and those that combine biobank and case-control samples using other methods, GPS better predicted disease progression.
Accurate prediction of disease progression using GPS can enable early interventions, targeted monitoring and personalized treatment decisions, leading to improved patient outcomes, Liu said. By identifying the people who are most likely to benefit from new therapies, it could also improve the recruitment and design of clinical trials.
A similar framework could be applied to studying other disease types, the researchers said, despite the focus of this study on autoimmune conditions.
” When we talk about underrepresented populations, it’s not just about race. It might also be a subset of patients who are understudied in the medical literature because they only make up a small portion of typical data sets. AI and transfer learning can aid in these populations ‘ studies and reducing health disparities, Liu said.
This research demonstrates the effectiveness of Penn State’s extensive research program in autoimmune disease.
Liu and Jiang— along with study co-authors , Laura Carrel, professor of biochemistry and molecular biology,  , Galen Foulke, associate professor of dermatology,  , Nancy Olsen, H. Thomas and Dorothy Willits Hallowell Chair in Rheumatology — formed the autoimmune working group and have collaborated for nearly a decade.
They conduct groundbreaking clinical trials, conduct research on the biological mechanisms of autoimmune diseases, and create AI techniques to address various issues relating to autoimmune diseases.
Chen Wang, who earned a doctorate in bioinformatics and genomics from Penn State, and Havell Markus, joint degree student in the , MD/PhD Medical Scientist Training Program, are co-first authors of the study.
Other Penn State authors on the paper include: Avantika R. Diwadkar, graduate student, Chachrit Khunsriraksakul, who graduated from the MD/PhD Medical Scientist Training Program during the time of the study, and Xingyan Wang, who was a research assistant at Penn State College of Medicine during the time of the study.
Other authors include Xue Zhong, research assistant professor of genetic medicine at Vanderbilt University School of Medicine, Xiaowei Zhan, associate professor of public health at the University of Texas Southwestern Medical Center, and Bingshan Li, professor of molecular physiology and biophysics.
Funding: Funding from the National Institutes of Health, including the National Institute of Allergy and Infectious Diseases Office of Data Science and Emerging Technologies, supported this research.
About this research in neurology and AI.
Author: Christine Yu
Source: Penn State
Contact: Christine Yu – Penn State
Image: The image is credited to Neuroscience News
Original Research: Open access.
Dajiang Liu et al.,” Integrating GWAS summary statistics and electronic health records to forecast the progression of autoimmune diseases from preclinical stages.” Nature Communications
Abstract
Integrating GWAS summary statistics and electronic health records to forecast the progression of autoimmune diseases from preclinical stages
Autoimmune diseases frequently exhibit a preclinical stage prior to diagnosis. Electronic health record ( EHR ) based-biobanks contain genetic data and diagnostic information, which can identify preclinical individuals at risk for progression.
In contrast to biobanks, which typically have a small number of cases and are insufficient to create precise polygenic risk scores ( PRS )? Importantly, progression and case-control phenotypes may have shared genetic basis, which we can exploit to improve prediction accuracy.
We propose a novel method , Genetic , Progression , Score ( GPS) that integrates biobank and case-control study to predict the disease progression risk. Prior to penalized regression, GPS incorporates PRS weights as prior and forces model parameters to be comparable to the prior to improve prediction accuracy.
In simulations, GPS consistently produces better prediction accuracy than alternative methods that only use biobank or case-control samples or those that combine both biobank and case-control samples.
The improvement is most apparent when the biobank sample is smaller or when there is a weak genetic correlation. In the BioVU biobank, we obtain PRS for the progression of systemic lupus erythematosus and preclinical rheumatoid arthritis, and we validate them in  , All of Us.
For both diseases, GPS achieves the highest prediction , R2 , and the resulting PRS yields the strongest correlation with progression prevalence.
