Regarding: LASSO-derived model for the prediction of lean-non-alcoholic fatty liver disease in examinees attending a routine health check-up

We have read with interest the recent paper titled ‘LASSO-derived model for the prediction of lean-non-alcoholic fatty liver disease in examinees attending a routine health check-up’ [1]. The authors present a prediction model constructed using the LASSO methodology to predict the risk of lean non-alcoholic fatty liver disease (lean-NAFLD) in a routine health check-up population. The researchers conducted a retrospective analysis of 2,325 subjects with a BMI < 24 kg/m². The sample was randomly divided into a training set and a validation set. Using the LASSO regression technique, the researchers screened 23 clinical and laboratory parameters and identified four key variables: visceral fat, triglycerides, high-density lipoprotein cholesterol (HDL-C), and waist-to-hip ratio. The model exhibited high discrimination in predicting lean-NAFLD with an AUC of 0.8416, comparable to the FLI model.

These results provide valuable insights for diagnosing and managing NAFLD and offer new theoretical support for further research in this area. However, the study could be improved in its design and implementation.

The study’s cross-sectional design and single-centre sample source may limit the generalizability of the results. To enhance the model’s generalizability, future studies should consider adopting a multicenter, prospective, long-term follow-up design and expanding the sample to include populations of different geographic regions and ethnicities.

Additionally, it is essential to note that the performance of the LASSO model is heavily dependent on the choice of regularization parameters [2,3]. The parameters included in this study may not fully cover all possible influences and may suffer from multicollinearity. To enhance the predictive accuracy and explanatory power of the model, it is recommended that researchers incorporate more biomarkers and genetic information in future studies. Furthermore, it is advisable to investigate methods for enhancing model performance by combining LASSO with other machine learning techniques, particularly in cases where complex feature interactions may prevent LASSO from identifying the global optimal solution.

Finally, The FLI has been demonstrated to be a valuable screening tool for NAFLD [4,5]. This study aimed to construct a large-scale prediction tool suitable for clinical settings. However, the proposed model’s prediction performance is comparable to the FLI, and the variables included, especially visceral fat, may be less accessible in practice than those included in the FLI. This may increase the difficulty of generalizing the model for clinical use. Researchers should consider investigating more cost-effective and convenient measurement methods for future work to enable the model’s application in a broader range of clinical practices.

In summary, this study presents a new perspective and method for predicting lean-NAFLD. However, further research is necessary to overcome existing limitations and realize its widespread application in clinical practice.

Authors contributions

Fan Zhang drafted the manuscript. Wenjian Li reviewed and approved the final version of this manuscript. All authors agreed to be accountable for all aspects of the work.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

The author(s) reported there is no funding associated with the work featured in this article.