AI/ML-enabled biosensing platforms for MASLD and MASH: Biomarkers, modalities, and translational readiness

Metabolic Dysfunction-Associated Steatotic Liver Disease (MASLD) and the progressive version Metabolic Dysfunction-Associated Steatohepatitis (MASH) are rising rapidly, yet most are still diagnosed late, because a liver biopsy is invasive and elastography and other specialist tools are not everywhere available. This is the reason why the time is now, since low-cost point-of-care testing (PoCT) technologies, smartphone-based optics, electrochemical sensors, lab-on-a-chip technologies, spectroscopy, breath/VOC, and Artificial Intelligence & Machine Learning (AI/ML) can take the noisy, variable signals produced by these technologies and make them reliable, repeatable signals. What is new in this article is that instead of discussing PoCT technologies as such, and instead of listing biomarkers in isolation, we are going to introduce the reader to the concept of a “biology → measurables → actions” framework in the context of MASLD/MASH management and explain how AI/ML needs to work throughout the entire PoCT process, from signal cleaning and automated quality control, to calibration transfer, uncertainty-based reporting, and decision support . This signal-to-decision approach distinguishes this review from the majority of existing articles, which focus mainly on the device's hardware, a single marker, or the accuracy of the most recent models but fail to consider the constraints of deployment. We will consider the major PoCT modality families of interest for NAFLD/NASH, including optical (colorimetry, fluorescence, chemiluminescence, electrochemiluminescence), electrochemical, microfluidics-based lab-on-chip, spectroscopy (Raman/SERS, NIR), and breath sensors/VOC sensor arrays. Clinicians, primary-care systems, researchers, and device developers will benefit from this roadmap to design safe, scalable MASLD/MASH screening and monitoring tools.