Hyderabad: A class XII student from Hyderabad, Ansh Kumar, has developed an artificial intelligence (AI)-assisted model for the early detection of lung cancer.Titled ‘DNA Methylation Biomarkers for Lung Cancer Detection: An AI-Driven Approach Using EGFR, PD-L1, SHOX2, RASSF1A and PTGER4’, the study analysed more than 7,000 patient samples and combined machine learning with DNA methylation biomarkers to improve non-invasive lung cancer detection.The work was done in collaboration with researchers from the Indian Council of Medical Research (ICMR) and researchers from the YRI Fellowship, USA, a programme that mentors young researchers in interdisciplinary science.According to the 17-year-old, the AI model aims to overcome some of the limitations of existing diagnostic methods, such as CT scans, biopsies and cytology, which may miss very early cancers, detect tumours only after they have advanced, produce false positives, or require invasive procedures.“Since DNA methylation changes can occur before tumours become visible, the AI system is designed to identify these subtle patterns and could potentially enable screening through blood plasma, sputum or bronchoalveolar lavage samples. By analysing multiple biomarkers simultaneously instead of relying on a single marker, the model seeks to improve sensitivity, specificity and risk prediction, potentially enabling earlier treatment,” Ansh said.His research has been accepted at the second International Conference on Information, Implementation and Innovation in Technology (IEEE).Supporting the potential of the approach, Muskan Modi, a researcher in the Microbiology Department of the Indian Council of Medical Research (ICMR), said there is strong evidence for biomarkers such as SHOX2 and RASSF1A methylation.“However, AI-specific prediction systems remain largely retrospective, small-scale and population-specific, meaning larger validation studies, reproducibility across populations and regulatory approvals are necessary before clinical adoption,” she said, adding that while it is a positive step, standardisation of DNA extraction methods, along with bioinformatics infrastructure, trained personnel and secure data storage, will be essential for widespread implementation.Speaking about AI in medical diagnostics, Dr Poornima Jogi from the YRI Fellowship, USA, said AI-assisted diagnostic pipelines can improve accessibility and reproducibility while also making decision-making more transparent by explaining why a prediction is made. “This will undergo multi-centre trials, prospective clinical testing and regulatory clearance before becoming part of routine hospital practice,” she added.
