AI is revolutionizing cancer care, enhancing diagnosis and treatment capabilities. At the AACR Annual Meeting 2025, researchers discussed AI’s potential to bridge gaps, especially in underserved areas. Notable advancements include the SMAART-AI model predicting cancer cachexia and pretrained models aiding in skin cancer diagnosis, indicating a promising future in personalized cancer treatment.
Artificial intelligence (AI) is making notable strides in the realm of cancer care, steadily proving itself as a transformative force. This was evident at the AACR Annual Meeting 2025 in Chicago, where brilliant minds gathered for a session entitled “Artificial Intelligence and Machine Learning for Basic and Translational Research.” Researchers there delved into how AI could help bridge significant gaps in cancer diagnosis and treatment, especially in under-resourced settings where technology is often scarce, and expert pathologists are hard to find.
Samantha Riesenfeld, a PhD candidate at the University of Chicago, co-moderated the session, alongside Paul Spellman from UCLA Health Jonsson Comprehensive Cancer Center. “This session highlights really exciting research…” she commented, emphasizing AI’s role in making predictions regarding diagnoses, subtyping cancers, targeted therapies, and prognoses. It’s a step towards better outcomes, especially when more accessible patient samples, like blood tests and pathology images, are factored in.
Take, for instance, Sabeen Ahmed, a graduate student at the University of South Florida. She unveiled an AI model named SMAART-AI at the conference, designed to predict cancer cachexia, a cruel symptom that leads to severe weight loss and muscle wasting. This condition severely affects pancreatic cancer patients, and, shockingly, it is linked to nearly 30% of cancer-related deaths.
Ahmed pointed out the urgency of early detection, as current methods often rely on subjective observations. “Detection of cancer cachexia enables lifestyle and pharmacological interventions…” she said, highlighting how crucial timely identification is. SMAART-AI utilizes imaging scans alongside routine clinical data to provide sophisticated predictions about cachexia risk, representing a novel application of AI to clinical biomarkers.
The process starts with the AI analyzing CT scans for muscle mass and then combines this with diverse clinical data, like lab results and patient history. Ahmed noted, “By combining all these data modalities, our AI-driven biomarker model…is designed to identify hidden patterns.” Remarkably, when lab results were included, the model’s accuracy soared to 85% in spotting cachexia among pancreatic cancer patients, a leap forward in personalized health strategies that could change lives.
Meanwhile, Steven Song, an MD/PhD candidate at the University of Chicago’s Pritzker School of Medicine, discussed AI’s application in diagnosing nonmelanoma skin cancer (NMSC). Song detailed how some regions, particularly those with arsenic-in-contaminated drinking water, struggle with expert resources for quick diagnosis of NMSC. His study looked into how pretrained AI models could step in where human expertise is lacking.
By using existing models, Song’s team demonstrated impressive results among over 2,000 pathology samples, achieving accuracies around 92.5% with the model named PRISM, which indicates a significant improvement over older architectures. “Our results demonstrate that pretrained machine learning models have the potential to aid diagnosis,” he contended.
It wasn’t just about performance, but accessibility too. Simplified AI models that require less complex data still showed robust capabilities, outperforming older systems. “Further work is needed to address practical considerations…” Song cautioned, reminding attendees of the hurdles left to overcome before implementing AI solutions on a larger scale.
The conversation didn’t end there. Both speakers underscored the emerging role of AI as a power player in cancer research and treatment. This ongoing transformation is ushering in a new era where personalizing care—tailoring treatments based on individual patient data—could lead to better outcomes than ever before. Keep an eye on AI, because the trajectory looks promising—as does the hope for countless patients.
Artificial intelligence is making waves in cancer care, from improving diagnostic accuracy to enhancing treatment approaches for conditions like cancer cachexia and skin cancer. The advances made at the AACR Annual Meeting showcase the potential of AI to fill gaps in under-resourced environments and even assist in personalizing patient treatment plans. As cutting-edge research continues to unfold, there is reason to believe that AI will play a pivotal role in the future of oncology, offering hope to many patients facing daunting diagnoses.
Original Source: www.aacr.org