Sifting through large quantities of data using artificial intelligence (AI) could one day change how nonalcoholic fatty liver disease (NAFLD) is diagnosed, managed, and treated, according to a recent review.
Artificial intelligence (AI) and its subsets may play a key role in the coming years in the diagnosis and management of the growing problem of nonalcoholic fatty liver disease (NAFLD), according to a recent review.
In particular, AI and its subdisciplines of supervised and unsupervised machine learning and deep learning and deep neural networks could assist with patient identification, determination of disease severity, and drug development in order to close knowledge gaps by sifting through vast amounts of data, the authors said.
Multiple studies have already identified several ways in which these advances—if verified with larger randomized controlled trials—could improve patient outcomes.
Applying deep learning to electronic health records could identify high-risk patients and clinical predictors of disease with improved accuracy. In patients with more advanced disease, deep learning and deep neural networks could reduce interobserver variability among pathologists, and, since the current gold standard for diagnosis is a liver biopsy, avoid the need for invasive testing. Earlier, easier identification can also enable lifestyle modifications, which are key to slowing disease progression.
As an example, the authors highlighted one study that used machine learning to predict the presence of NAFLD in the general population with MRI imaging available in order to create a model that produced the NAFLD ridge score, which included 6 clinical characteristics (alanine aminotransferase, high-density lipoprotein cholesterol, triglyceride, hemoglobin A1c, white blood cell count, hypertension). In the training cohort, the area under the receiver operating characteristic curve was 0.87 and in the validation cohort it was 0.88.
In another study involving more than 10,000 patients, researchers used AI to identify triglycerides, alanine aminotransferase, gamma-glutamyltransferase, uric acid, and body mass index as the most effective predictive markers of NAFLD.
In stratifying severity in patients with NAFLD as well as nonalcoholic steatohepatitis and advanced fibrosis, the authors highlighted a research platform that was used to train a deep neural network with more than 20 layers and 8 million parameters from more than 600 liver biopsies. The machine learning models and the pathologist interpretations had strong agreement, which could pave the way in the future to improved work efficiency, increase uniformity, and speed processing.
As there are limited options for NAFLD drug therapy, the authors said, AI could identify patient response, particularly by taking into account their genomics and environmental exposures. AI could also be used to identify possible new drug targets or repurpose existing medications.
Reference
Aggarwal P, Alkhouri N. Artificial intelligence in nonalcoholic fatty liver disease: A new frontier in diagnosis and treatment. Clin Liver Dis. Published online August 5, 2021. doi:10.1002/cld.1071
What We’re Reading: ChatGPT in Health Care; Menthol Ban Meeting; Health System Cyberattack Impact
November 30th 2023ChatGPT is changing health care while also raising questions about artificial intelligence's promises and limitations; 24 Biden officials met with the National Organization of Black Law Enforcement Executives to discuss the FDA’s proposal to ban menthol cigarettes; a recent hospital cyberattack shows how the vulnerability of health care systems can put patients at risk.
Read More
Emily Goldberg Shares Insights as a Genetic Counselor for Breast Cancer Risk Screening
October 30th 2023On this episode of Managed Care Cast, Emily Goldberg, MS, CGC, a genetic counselor at JScreen, breaks down how genetic screening for breast cancer works and why it is so important to increase awareness and education around these screening tools available to patients who may be at risk for cancer.
Listen
Study Validates Smartphone Analysis for Predicting Obstructive Sleep Apnea
November 29th 2023Findings show breathing sounds recorded by smartphones during sleep at home can reliably identify obstructive sleep apnea, presenting an innovative and accessible alternative to traditional in-laboratory polysomnography.
Read More
Examining Telehealth Uptake to Increase Equitable Care Access
January 26th 2023To mark the publication of The American Journal of Managed Care®’s 12th annual health IT issue, on this episode of Managed Care Cast, we speak with Christopher M. Whaley, PhD, health care economist at the RAND Corporation, who focuses on health economics issues, including the influence of the COVID-19 pandemic on health care delivery.
Listen
Researchers Identify Potential Biomarkers for SMA Motor Improvement Following Nusinersen Treatment
November 17th 2023A retrospective, exploratory analysis identifies biomarkers predictive of spinal muscular atrophy (SMA) response to nusinersen and suggests the efficacy of using machine learning algorithms to anticipate patient outcomes.
Read More
Data on Delandistrogene Moxeparvovec Can Inform Future Gene Therapies
November 17th 2023The FDA approved delandistrogene moxeparvovec-rokl (Elevidys, Sarepta Therapeutics) in June to treat Duchenne muscular dystrophy in ambulatory pediatric patients aged 4 through 5 who have a confirmed DMD gene mutation.
Read More