Machine learning (ML) can generate clinically meaningful data regarding liver disease outcomes in patients with alpha-1 antitrypsin deficiency (AATD), according to a study published in Scientific Reports. 

AATD often leads to pulmonary and liver disease, with the severity of clinical outcomes dependent largely on the patient genotype. Individuals with the PiSZ and PiMZ genotypes tend to have a milder disease course, whereas those with the more common PiZZ genotype tend to have poorer prognosis. 

Studies indicate that liver damage can progress insidiously for decades before any symptoms are observed. The course of liver disease in patients with AATD is accelerated by other factors such as excessive alcohol consumption, smoking, and nonalcoholic fatty liver disease. It is vital that patients who are likely to develop severe liver disease are identified for timely intervention to take place. 


Continue Reading

“Although ML techniques have been applied in the medical field for disease diagnosis and treatment outcome prediction, given recent advances in ML algorithms and statistical computing power, little research has been done to understand the AATD-liver disease patients’ journey or predict the disease progression of liver disease associated with AATD using ML algorithms,” the authors wrote. 

Read more about AATD etiology 

The researchers established a predictive ML model to assess the progression of liver disease in patients with AATD using real-world data from the UK Biobank. In addition, they sought to improve their ML model’s predictive accuracy by combining various ML algorithms and using artificial neutral network multilayer perception via meta-learning. 

They reported that the predictive ML model correlated with existing studies on AATD liver progression in 4 key outcomes: all-cause mortality, liver-related death, liver transplant, and all-cause mortality/liver transplant. The UK Biobank provided a wealth of data that allowed researchers to build a ML predictive model with clinically meaningful accuracy. 

“Our work may lead to greater insights in clinical practice and assist clinicians in effectively identifying high-risk patients with liver disease associated with AATD, mitigating the burden of diagnosis and in managing the disease progression and treatment,” the authors wrote. 

Reference

Meng L, Treem W, Heap GA, Chen J. A stacking ensemble machine learning model to predict alpha-1 antitrypsin deficiency-associated liver disease clinical outcomes based on UK Biobank dataSci Rep. 2022;12(1):17001. doi:10.1038/s41598-022-21389-9