Comment on “GLM7 – A Novel Composite Glycolipid Index Derived from Routine Health Indicators for Enhanced Diagnosis and Prediction of Multimorbidity”

Article to which Correspondence refers https://doi.org/10.1002/advs.202510552 Wang Z, Chen S, Feng X, Chen X, Evans PC, Strijdom H, et al. GLM7 – A Novel Composite Glycolipid Index Derived from Routine Health Indicators for Enhanced Diagnosis and Prediction of Multimorbidity. Adv Sci (Weinh). 2025, 12, 10552. Recently, Zhihua Wang and his colleagues [1] developed GLM7 (Glycolipid Metabolism 7 Factors), a novel disease prediction index. They investigated its predictive capability across multiple common diseases. GLM7 incorporates seven standard indicators, including fasting blood glucose (FBG), insulin, and BMI. The study's findings reveal that GLM7 has predictive value for diseases such as tumors, cardiovascular disease, and diabetes. The study also preliminarily defines the risk threshold range for GLM7. Furthermore, they validated the predictive model's performance in two independent cohorts in China and the United States. Using the NHANES database, the authors conducted their research and validated their findings using the CHARLS database. This approach overcomes the limitations of single-disease and single-population studies. Together, these results demonstrate the potential of GLM7 for precision medicine and offer promising prospects for personalized healthcare. Routine health examinations play an irreplaceable role in the prevention and early diagnosis of various diseases, including cancer, cardiovascular diseases, and diabetes. Certain indicators enable the early identification of diseases, thereby reducing mortality rates, improving prognosis, and optimizing public health strategies. For instance, in oncology, research indicates that neuron-specific enolase (NSE) serves as a key biomarker for small cell lung cancer (SCLC). Frequently utilized in clinical diagnosis, it contributes to enhancing the efficiency of early detection [2, 3]. Additionally, studies have demonstrated that the combination of carcinoembryonic antigen (CEA) with markers such as CA125 can be utilized for monitoring treatment efficacy and assessing prognosis in advanced lung adenocarcinoma [4, 5]. However, these specific tumor marker tests are often costly, and cancer patients tend to be older, frequently presenting with comorbidities such as hypertension, coronary heart disease, and diabetes. Currently, there is a need for efficient, affordable, and easily measurable indicators to predict the associated risks of these common diseases. Against this backdrop, Zhihua Wang and colleagues developed a novel composite indicator, GLM7, which integrates multiple baseline physical examination metrics and demonstrates promising predictive value for various diseases and comorbidities. This holds significant implications for public health and clinical practice. After carefully reviewing the article, we wish to discuss several issues with experts in oncology, endocrinology, and cardiovascular medicine, as well as the authors of this paper. First, the study defines diabetes as having a history of diabetes or a glycated hemoglobin HbA1c level greater than 6.5%. It is well known that HbA1c is considered the gold standard for diabetes diagnosis. However, blood glucose concentration estimation is also a critical diagnostic tool, typically involving FBG testing or an oral glucose tolerance test (OGTT) to detect whether blood glucose levels exceed the threshold [6, 7]. The GLM7 indicator constructed in this study already includes the FBG metric. The conceptual proximity between the predictor and the outcome may lead to an overestimation of the index's apparent predictive performance, which complicates the interpretation of its clinical utility. The use of GLM7 to explore associations with diabetes may warrant further discussion. Additionally, the study defines liver disease as any liver condition diagnosed by a physician. However, in clinical practice, liver disease diagnosis typically relies on imaging studies and liver biopsy [8, 9]. The predictive performance of the GLM7 index for liver disease in this study may require further investigation in clinical applications. Second, from a methodological standpoint, the work underscores the importance of construct validity and scale handling when designing composite indices like GLM7. The study presents logistic regression analyses suggesting that GLM7 is a strong risk factor for several outcomes, while age and some traditional risk factors appear to have comparatively modest effect estimates. Such large contrasts in estimated risk contributions may, in part, reflect how GLM7 was mathematically constructed and scaled, rather than purely biological differences in causal importance. We also found no mention of automatic data normalization or standardization in the authors' methodological description. This may indicate a potential dimensional-analysis trap, which warrants further investigation. Third, we were unable to locate insulin-related statistics directly within the CHARLS database. Since accessing CHARLS data requires an email application, unlike the NHANES database, which allows straightforward downloads, could the authors please advise how insulin-related data can be obtained directly from CHARLS, or through which indicators it can be converted? Finally, the statistical reporting in the study raises a more general consideration regarding robustness, multiple testing, and internal consistency of the results. The authors evaluate GLM7 in relation to multiple outcomes and covariates, using a range of regression models and performance metrics. In such multidimensional analyses, careful control of type I error, transparent handling of multiple comparisons, and consistency between point estimates, confidence intervals, and p‑values are essential to avoid overinterpreting marginal associations. Although the study provides a rich set of statistical outputs, future work could further strengthen the credibility of GLM7 by explicitly detailing the strategy for multiple testing adjustment, reporting effect sizes alongside absolute risk differences, and ensuring that statistical inferences are aligned across all reported metrics. Such refinements would not only improve the methodological rigor of GLM7-related research but also facilitate its acceptance by clinicians, statisticians, and policymakers. In summary, early disease detection and risk prediction are still significant challenges in medicine. We sincerely commend the authors for their innovative work introducing the novel composite indicator, GLM7. This development opens new avenues for researching the diagnosis and prediction of common diseases, such as cancer and diabetes. As follow-up studies continue, we believe this research will greatly advance precision medicine. The authors declare no conflicts of interest. Data sharing not applicable to this article as no datasets were generated or analysed during the current study.