Advances in AI-driven diagnostics from the AMP 2025 conference
At AMP 2025, experts present groundbreaking AI applications in molecular diagnostics, including high-accuracy cancer classifiers, noninvasive CNS tumor detection via cerebrospinal fluid, and AI-assisted cytogenetics. These innovations aim to improve diagnostic precision and support personalized medicine.
Within the field of molecular pathology, Artificial intelligence is being used in part to improve diagnostic processes and accuracy. AI has the potential to not only automate tasks, but also to enhance clinical decision making.
Innovative studies in diagnostic applications of artificial intelligence will be presented at the Association for Molecular Pathology (AMP) 2025 Annual Meeting & Expo, taking place Nov. 11–15 in Boston.
Below are some of the potential AI applications that will be shared by leading experts in molecular diagnostics at AMP 2025:
93% diagnostic accuracy achieved by AI classifier for cancer: Researchers from The Hospital for Sick Children developed a web platform incorporating an AI classifier designed to handle heterogeneous datasets and integrate RNA sequencing into clinical workflows. The model achieved 93% diagnostic accuracy on subtypes covered by the platform. The system can also adapt and incorporate new subtypes, increasing its accuracy with each new sample. The goal for the platform is to cover new subtypes with only five reference samples.
Earlier, noninvasive diagnosis using spinal fluid enabled by AI: Central nervous system tumor tissue biopsies are invasive and difficult to repeat, limiting their role in diagnosis. Cerebrospinal fluid–derived circulating tumor DNA offers a noninvasive alternative. Researchers at Soonchunhyang University in South Korea created two AI models to classify samples: a dense neural network trained on mutation data from 12 key genes via next-generation sequencing and a convolutional neural network trained on standardized MRI images. Both models showed strong accuracy, the researchers said. Combining outputs from both models improved prediction and classification accuracy.
AI reveals chromosomal changes in blood cancer patients: Researchers at Wake Forest University School of Medicine used an AI-trained karyotyping algorithm in clinical cytogenetics to analyze chromosomal abnormalities in GATA2 deficiency syndrome-related leukemia. AI enabled rapid generation and review of hundreds of images, improving detection and confidence in identifying complex clonal chromosome rearrangements. The AI-assisted karyotyping revealed detailed clonal evolution over time in a patient’s acute myeloid leukemia, capturing multiple chromosomal changes that tracked disease progression. Overall, the algorithm provided insights into personalized disease progression and better understanding of GATA2 deficiency syndrome.
With AI, doctors move toward more personalized oncology care: At Augusta University, researchers created a computational framework to train and compare AI models to look at hematoxylin and eosin (H&E)-stained slide images from patients and predict genomic and transcriptomic information directly from the images. The framework successfully compared different AI models and could handle prediction tasks such as identifying gene activity and prognosis. When patient clinical information was incorporated, the predictions became more informative. The researchers found that different AI models perform better or worse depending on the specific goal or data type, requiring future standardization of testing and benchmarks.
