Machine Learning to Identify Metabolic Subtypes of Obesity: A Multi-Center Study

Abstract

Background and objectiveClinical characteristics of obesity are heterogenous, but current classification for diagnosis is simply based on BMI or metabolic healthiness. The purpose of this study was to use machine learning to explore a more precise classification of obesity subgroups towards informing individualized therapy.Subjects and MethodsIn a multi-center study (n=2495), we used unsupervised machine learning to cluster patients with obesity from Shanghai Tenth People’s hospital (n=882, main cohort) based on three clinical variables (AUCs of glucose and of insulin during OGTT, and uric acid). Verification of the clustering was performed in three independent cohorts from external hospitals in China (n = 130, 137, and 289, respectively). Statistics of a healthy normal-weight cohort (n=1057) were measured as controls.ResultsMachine learning revealed four stable metabolic different obese clusters on each cohort. Metabolic healthy obesity (MHO, 44% patients) was characterized by a relatively healthy-metabolic status with lowest incidents of comorbidities. Hypermetabolic obesity-hyperuricemia (HMO-U, 33% patients) was characterized by extremely high uric acid and a large increased incidence of hyperuricemia (adjusted odds ratio [AOR] 73.67 to MHO, 95%CI 35.46-153.06). Hypermetabolic obesity-hyperinsulinemia (HMO-I, 8% patients) was distinguished by overcompensated insulin secretion and a large increased incidence of polycystic ovary syndrome (AOR 14.44 to MHO, 95%CI 1.75-118.99). Hypometabolic obesity (LMO, 15% patients) was characterized by extremely high glucose, decompensated insulin secretion, and the worst glucolipid metabolism (diabetes: AOR 105.85 to MHO, 95%CI 42.00-266.74; metabolic syndrome: AOR 13.50 to MHO, 95%CI 7.34-24.83). The assignment of patients in the verification cohorts to the main model showed a mean accuracy of 0.941 in all clusters.ConclusionMachine learning automatically identified four subtypes of obesity in terms of clinical characteristics on four independent patient cohorts. This proof-of-concept study provided evidence that precise diagnosis of obesity is feasible to potentially guide therapeutic planning and decisions for different subtypes of obesity.Clinical Trial Registrationwww.ClinicalTrials.gov, NCT04282837.

A multi-center study on glucometabolic response to bariatric surgery for different subtypes of obesity

“…Table 1 shows the anthropometry characteristics of each AIM subtype of obesity patients at baseline and 12-month postsurgery. Patients in the HMO-U and HMO-I subgroups were characterized by higher pre-surgery weight and BMI, consistent with what was observed in our previous study (11). During the 12-month time period after surgery, patients in both the HMO-U and HMO-I subgroups showed the most weight loss (33.5% and 34.0%, respectively) and WC loss (23.9% and 26.7%, respectively), while patients in the LMO group showed the least weight loss (24.4%) and WC loss (18.2%).…”

“…At baseline (Figure 3A), 54.4% of patients with obesity had hyperinsulinemia, in which patients in HMO-I showed the highest rate (87.8% vs. 45.1-66.7% in the other three subgroups), consistent with what we reported previously (11).…”

The potential of DNA methylation as a biomarker for obesity and smoking

“…However, the underlying molecular mechanisms remain poorly characterized as the research is hampered by the multifactorial nature of obesity. Obese individuals may display highly dissimilar metabolic characteristics from each other [69]. Hence, it seems that excess weight does not solely explain why some people are more prone to develop related diseases.…”

“Big Data” Approaches for Prevention of the Metabolic Syndrome

“…Indeed, many studies of machine learning in metabolic disorders, including MetS, type II diabetes, and hypertension, focus on mining data from extensive medical documentation such as electronic records and medical images from daily practices and large-scale trials ( Lin et al, 2021b ; Yu et al, 2021 ). Common data types would include anthropometric data, laboratory tests covering different systems, and medical imaging of other modalities such as x-ray based, ultrasound-based, and even MRI-based techniques.…”