With the number of pathologists declining, team develops AI to be used as a diagnostic tool NatureComms
Fig. 2: Convolutional neural networks achieve high performance in the prediction of PD-L1 and PD-1 expression.
Receiver operating characteristics curves for the performance of the proposed models, in terms of AUC, for PD-L1 and PD-1 prediction in the BCCA and MA31 cohorts.The model obtained high prediction accuracies for both the BCCA cross-validation and BCCA test set . When analyzing only concordant cases between pathologists, AUC performance was further increased .For the external MA31 cohort, the performance dropped to 0.
To compensate for these cohort differences and calibrate the model to better fit MA31, we applied a transfer learning approachto fine-tune only the last layer of the CNN using the TMAs from some of the patients . The patients whose TMAs were used for fine-tuning the model were excluded from the inference and prediction analysis. The calibration resulted in prediction scores that obtained an AUC performance of 0.886 for MA31 , which was higher by 0.036 than the AUC of the uncalibrated system.
We set a threshold, such that patients who obtained prediction scores below or above that threshold were classified as low-prediction score or high-prediction score , respectively. Low-PS patients were predicted as PD-L1 negative, and high-PS patients were predicted as PD-L1 positive. The threshold was tuned once using the BCCA cross-validation to obtain a large low-PS group while maintaining high sensitivity .
France Dernières Nouvelles, France Actualités
Similar News:Vous pouvez également lire des articles d'actualité similaires à celui-ci que nous avons collectés auprès d'autres sources d'information.
Metagenomics of Parkinson’s disease implicates the gut microbiome in multiple disease mechanisms - Nature CommunicationsHere, the authors perform large-scale high-resolution Parkinson’s disease metagenomics analyses, revealing widespread dysbiosis characterized by overabundance of pathogens, immunogens, toxicants, and curli, reduction in neuroprotective and antiinflammatory molecules, and dysregulated neuroactive signaling.
Lire la suite »
A self-powered ingestible wireless biosensing system for real-time in situ monitoring of gastrointestinal tract metabolites - Nature CommunicationsMetabolic dynamics within the small intestine are difficult to study due to the lack of in situ access. Here, the authors report an ingestible, self-powered, and wireless biosensing system, demonstrating proof-of-principle real-time glucose monitoring in the small intestines of pigs.
Lire la suite »
Defective activation and regulation of type I interferon immunity is associated with increasing COVID-19 severity - Nature CommunicationsThe interferon response has been shown to be linked to severity of SARS-CoV-2 infection and is an essential component of the immune response to COVID-19. Here the authors stratify patients according to COVID-19 severity and asses the interferon response showing defective responses in severe infection and highlight the importance of assay variables and confounding factors that impact the detection of interferon.
Lire la suite »
PLD3 affects axonal spheroids and network defects in Alzheimer’s disease - NatureAmyloid-plaque-associated axonal spheroids are prominent contributors to neural network dysfunction in an Alzheimer’s model and can be reversed by endolysosomal modulation.
Lire la suite »
Ketamine triggers a switch in excitatory neuronal activity across neocortex - Nature NeuroscienceCichon et al. show that ketamine induces a switch in the active population of excitatory neurons across cortical layers and regions that contributes to impairments in sensory processing characteristic of a dissociative-like state.
Lire la suite »