Researchers from the University of Missouri used artificial intelligence (AI) to analyse the publicly available data from about 16,000 participants enrolled in the T1D Exchange Registry database to learn more about people with Type 1 diabetes.
The information was acquired by the team using health informatics in order to gain a better understanding of the disease, and they were sponsored in part by a grant from the National Science Foundation in the United States.
According to Chi-Ren Shyu, one of the researchers, they delegate the task of connecting millions of dots in the data to the computer so that it could identify major contrasting patterns between people who had a family history of Type 1 diabetes and people who did not have such a history. Additionally, they had the computer perform the statistical testing so that they could be sure of the accuracy of their findings.
“We let the computer do the statistical testing to make sure we are confident in our results,” said Shyu.
The investigation carried out by the researchers came up with a few results that were surprising. Citing an example when they discovered that people who had an immediate family member with type 1 diabetes were more likely to be diagnosed with hypertension, in addition to diabetes-related nerve disease, eye disease, and kidney disease.
This information was gleaned from the registry of people with diabetes. Erin Tallon, who was the principal author of the study, said that the findings prove that real-world data and artificial intelligence have utility.
Using AI, researchers are attempting to discover a solution to the issue that persons with Type 1 diabetes would manifest the symptoms in various ways. Diabetes Type 1 is not a distinct disease that presents itself in the same way for everyone who has it. By doing analysis on data gathered from the real world, they can gain a better knowledge of the risk variables that may put a person at a higher risk for developing bad health outcomes.
Researchers are hoping that their discoveries can help solve a more widespread issue by employing more extensive data sets that are population-based. They are also planning to construct larger patient cohorts, do additional data analysis, and make use of these algorithms or artificial intelligence to assist them.
An algorithm for mining contrast patterns can identify statistically significant deviations in the distribution of attribute frequencies between two patient categories. The validated approach was utilised by the researchers to find individual and co-occurring traits that were observed considerably more frequently in familial Type 1 diabetes compared to sporadic cases of Type 1 diabetes.
They talk about these traits using the term “patterns” or “feature patterns.” While the method returns feature patterns with one, two, or three components, these patterns might have any number of elements. The terms “individual elements” and “individual characteristics” are interchangeable.
This study is the largest one done to date to compare the health outcomes of patients with familial and sporadic types of Type 1 diabetes over a longer period of time. It was conducted using data from the T1D Exchange Clinic Registry with thousands of participants.
The current findings need to be validated in a larger population-based cohort by the conduct of additional research. This approach has the potential to be modified to assist in the creation of individualised treatment options for diabetic patients.