MD Anderson researchers developed the CIPHER AI model to predict pneumonitis in lung cancer patients using baseline CT scans, outperforming existing models.
Key Details
- 1CIPHER is a deep learning AI model based on CT imaging for lung cancer patients starting immunotherapy.
- 2The model was trained on over 590,000 CT slices from 2,500 patients.
- 3Internal validation involved 93 patients, while external validation used a cohort of 116 patients from Johns Hopkins.
- 4CIPHER achieved AUC values between 0.77 and 0.85, outperforming clinical and radiomics models.
- 5In external validation, CIPHER had an AUC of 0.83 and balanced accuracy of 81.7%, with higher specificity than the radiomics comparator.
- 6Prospective studies are still needed for broader clinical implementation.
Why It Matters
Identifying patients at risk for pneumonitis before immunotherapy could enable early intervention and tailored monitoring, potentially improving outcomes and patient safety in oncologic care. This AI model exemplifies the potential of imaging-based predictive analytics in personalizing cancer treatment.
Source
AuntMinnie
Related News
•Radiology Business
LLMs Match Radiologists in MRI Protocol Selection Accuracy
Large language models can match radiologists in selecting MRI protocols, potentially improving workflow efficiency.
•Radiology Business
Weill Cornell Investigates Unauthorized Use of Radiology Data in AI Research
Weill Cornell Medicine reported improper use of patient information in radiology AI research, violating privacy protocols.
•AuntMinnie
Open-Source AI Platform Powers Pediatric Bone and Fetal MRI Pipelines
Open-source AI platform successfully integrated and sustained two pediatric imaging pipelines for bone length and fetal MRI at a children's hospital.