Posts

Why Most Clinical AI Studies Disappear: A Successful Colorectal Surgery Prediction Model Author: Bruno Pignataro

Image
 Author: Bruno Pignataro TL;DR A Danish AI model predicting 1-year mortality after colorectal cancer surgery was successfully deployed in clinical practice — rare for clinical AI. Patients receiving AI-guided care had 36% fewer postoperative complications compared to standard care. Cost savings reached $2,847 per patient in the first year. Success relied on a high-quality national dataset, surgeon involvement, and prediction linked directly to treatment. ‍ Most clinical AI models stop at performance metrics. Here’s a Danish colorectal surgery model that made it into practice, and what we can learn from it. AI-based tools for personalized clinical decisions are still largely limited to research settings. We face substantial complexity in both building and implementing these models. Studies often develop a model, report an area under the curve (AUC), and then simply disappear. This week, I came across an interesting article. Andreas Weinberger Rosen and his Danish co-authors develop...

Imaging-Driven Real-World Evidence in Oncology: Advancing Precision Medicine Through Disease-Specific Research

Image
Clinical trials have been the foundation of how we understand the field of oncology for many years. However there is one fundamental flaw in how we understand oncology through trials. Historically, the overall estimated patient participation rate to cancer treatment trials was 7.1%.At the community level, where the majority of patients are treated, only about 4% of adult cancer patients are treated in research trials. This evidence gap is incredibly significant compared to the total number of oncology patients in the real-world. Most of the patients treated for cancer are getting treated in community-based programs and not included in any kind of formal research documentation.‍ One of the ways to address this problem is with real-world evidence (RWE) . Real-world evidence is documented clinical evidence that comes from routine healthcare and not clinical research settings. Real-world evidence provides a clear understanding of how real patients receive treatment in real hospitals. When ...

Single-Center vs Multi-Center Studies: What We’ve Learned About Scaling Imaging AI | Segmed Experts

Image
‍From Proof-of-Concept to Scale‍ Over the past ten years, artificial intelligence (AI) has progressed from research-oriented prototypes to clinical decision support systems in medical imaging. The U.S. Food and Drug Administration (FDA) has approved more than a thousand AI/ML-assisted medical devices , the majority of which are imaging-based. The early successes of imaging AI often begin in the highly resourceful academic setting. A research team, often affiliated with a leading technical university, collaborates with a renowned academic medical center. The purpose of this collaboration is to leverage a carefully curated retrospective dataset precisely matching the target application of the AI model. These datasets usually have structured reporting, uniform imaging protocols, stable scanner configurations, and relatively complete metadata. In such controlled settings, model development is rapid as datasets are easily standardized and clinical outcomes are well-defined. Internal validat...