As Geneia blog readers know, I am very proud of the Geneia Data Intelligence Lab (GDI Lab) and the important work the Ph.D.- and masters-level data scientists do to solve healthcare’s intractable questions such as who are the most impactable patients.
In March, I introduced Zhipeng Liu to the Geneia blog community. He’s one of the lab’s principal data scientists.
To refresh your memory, Zhipeng holds a Master of Science in statistics and a Ph.D. in pharmacology from Purdue University. During his five years at Purdue, he studied how genetics factors contribute to the development of metabolic diseases such as fatty liver and type 2 diabetes, and published six peer-reviewed papers in high-impact journals. As a data science fellow, he created a predictive model that uses data in the EHR to identify people at high risk for developing fatty liver diseases. Zhipeng also has a master’s in biochemistry and bachelor of engineering in bioengineering.
I am pleased to share that part of Zhipeng’s Ph.D. work was published in the August issue of the highly-ranked Journal of Hepatology. In the article, Causal relationships between NAFLD, T2D and obesity have implications for disease subphenotyping, Zhipeng and his co-authors explored the relationships between non-alcoholic fatty liver disease (NAFLD), type 2 diabetes (T2D) and obesity from the perspective of genetics and provided insights for disease subphenotyping.
In other words,
“Non-alcoholic fatty liver disease, type 2 diabetes and obesity are epidemiologically correlated with each other, but their causal relationships were incompletely understood. Herein, we identified causal relationships between these conditions, which suggest that each of these closely related diseases should be further stratified into subtypes. This is important for accurate diagnosis, prevention and treatment of these diseases.”
Q&A with Zhipeng Liu
I had the opportunity to ask Zhipeng some questions about his research, which were:
What were the major findings and significance of your study?
NAFLD, T2D, and obesity are often presented together in patients and treated as a comorbidity with each other. However, without a clear understanding of their causal relationships, drug development and treatment strategy may fail because the underlying root cause is unknown. This study, for the first time, explored the causal relationships between these three important metabolic diseases and implied novel subtypes of the diseases.
In brief, the research discovered that NAFLD can be separated into at least two subtypes: those mainly caused by ‘nature’ (genetics) and those caused by ‘nurture’ (metabolic disorders such as T2D or obesity). Genetically-driven NAFLD can promote the development of hyperglycemia, but not necessarily insulin resistance.
Surprisingly, NAFLD does not lead to overall obesity, but promotes the development of central obesity. In the ‘nurture' model, both T2D and obesity can promote NAFLD. In this case, NAFLD is secondary to T2D or obesity.
How does your study help patients and physicians?
This study provides important insights into disease classification, diagnosis and treatment. For example, patients with genetically-driven NAFLD are likely underdiagnosed since their tendency to be lean and less resistant to insulin makes them appear to be healthy. For the patients with this disease subtype, treatment should be focused on targeting the genetic causes in the liver. On the other hand, people with NAFLD that is driven by metabolic disorders, e.g. T2D and obesity, are likely to benefit from weight reduction and blood glucose control.
How does your study help health plans?
The study suggests health plans should offer more genetic tests and routine health screening, which will help with early diagnosis and intervention and ultimately reduce medical costs.
Is there any connection between this research and the work you’re doing at Geneia?
At Geneia, I am focused on improving the outcomes and reducing the costs of chronic diseases through predictive modelling. A better understanding of the heterogeneity of chronic disease will be helpful for developing more accurate models and making more specific suggestions for care management.
What is the current focus of your work at Geneia?
Since I joined Geneia, I’ve been working to improve the high-cost claimants model. The high-cost claimants model uses claims data from the previous 12 months to predict the future cost in the next 12 months. The latest version of the high-cost model can precisely predict future cost using demographics information (e.g., age and gender) and as few as only two data elements from claims (e.g., cost and date of service).
Next, I will be refining the type 2 diabetes onset prediction model.
To learn more about the Geneia Data Intelligence Lab and the work of Geneia’s data scientists, click here.