A rare disease is one that affects fewer than 200,000 individuals. Despite their rarity, over 10,000 known rare diseases collectively impact more than 30 million people. While some diseases like multiple sclerosis are well-recognized, most—including bartonellosis, maple syrup urine disease, and visual snow syndrome—remain obscure. Acute hepatic porphyria (AHP), a group of rare genetic disorders, affects about 1 in 100,000 people, predominantly women. Symptoms often correlate with the menstrual cycle and can lead to severe, life-threatening episodes characterized by abdominal pain, nausea, vomiting, limb weakness, and anxiety. Although the FDA approved givosiran in 2019 as a prophylactic treatment, diagnosing AHP typically takes 15 years due to its rarity and the infrequency with which physicians encounter it. During this delay, the disease can worsen and cause irreversible damage.
In a groundbreaking development, researchers have created an artificial intelligence (AI) tool that can identify potential AHP patients, aiding in the diagnosis and management of this rare genetic disease. Developed by UCLA Health under Project Zebra, this predictive algorithm sifts through electronic health records to spot patterns indicative of the disease, alerting physicians to patients who may need further testing and diagnosis. The algorithm, trained on a dataset from UCSF and UCLA, analyzes de-identified patient records from approximately 10 million records throughout the UC system. The development faced challenges, such as the lengthy approval process for accessing patient data and the need for vast datasets to improve the algorithm’s accuracy. To overcome this, zebraMD implemented Virtual Pooling, a patented technology allowing the algorithm to learn from data without transferring it.
One significant hurdle in developing the algorithm was the unorganized nature of patient data, which includes both structured data like vital signs and lab results and unstructured data such as physician’s notes. The latter presents particular difficulties for algorithms, prompting researchers to refine and structure the data. The model was equipped with detailed symptom information and references from a rare and genetic disease database maintained by the National Institutes of Health. The algorithm independently scanned the data to identify patterns, a crucial approach given the high rate of misdiagnosis associated with rare diseases.
According to findings published in the Journal of the American Medical Informatics Association, the algorithm accurately predicted referrals for AHP testing with an accuracy of 89% to 93%, and it identified 71% of patients who tested positive for the disease earlier than their actual diagnosis, saving an average of 1.2 years per case. Project Zebra is now expanding its focus to include predicting cerebral aneurysms, a potentially life-threatening condition.
Article written by HospiMedica
05/07/2024
Source:
HospiMedica
https://www.uoc.edu/portal/es/ehealth-center/actualitat/