Development of an AI-based support system for controlled ovarian stimulation.

Purpose: Controlled ovarian stimulation (COS) is vital for IVF. We have developed an AI system to support the implementation of COS protocols in our clinical group. Methods: We developed two models as AI algorithms of the AI system. One was the oocyte retrieval decision model, to determine the timing of oocyte retrieval, and the other was the prescription inference model, to provide a prescription similar to that of an expert physician. Data was obtained from IVF treatment records from the In Vitro Fertilization (IVF) management system at the Asada Ladies Clinic, and these models were trained with this data. Results: The oocyte retrieval decision model achieved superior sensitivity and specificity with 0.964 area under the curve (AUC). The prescription inference model achieved an AUC value of 0.948. Four models, namely the hCG prediction model, the hMG prediction model, the Cetrorelix prediction model, and the Estradiol prediction model included in the prescription inference model, achieved AUC values of 0.914, 0.937, 0.966, and 0.976, respectively. Conclusion: The AI algorithm achieved high accuracy and was confirmed to be useful. The AI system has now been implemented as a COS tool in our clinical group for self-funded treatments.

Deep learning and digital pathology powers prediction of HCC development in steatotic liver disease.

BACKGROUND AND AIMS: Identifying patients with steatotic liver disease who are at a high risk of developing HCC remains challenging. We present a deep learning (DL) model to predict HCC development using hematoxylin and eosin-stained whole-slide images of biopsy-proven steatotic liver disease. APPROACH AND RESULTS: We included 639 patients who did not develop HCC for ≥7 years after biopsy (non-HCC class) and 46 patients who developed HCC <7 years after biopsy (HCC class). Paired cases of the HCC and non-HCC classes matched by biopsy date and institution were used for training, and the remaining nonpaired cases were used for validation. The DL model was trained using deep convolutional neural networks with 28,000 image tiles cropped from whole-slide images of the paired cases, with an accuracy of 81.0% and an AUC of 0.80 for predicting HCC development. Validation using the nonpaired cases also demonstrated a good accuracy of 82.3% and an AUC of 0.84. These results were comparable to the predictive ability of logistic regression model using fibrosis stage. Notably, the DL model also detected the cases of HCC development in patients with mild fibrosis. The saliency maps generated by the DL model highlighted various pathological features associated with HCC development, including nuclear atypia, hepatocytes with a high nuclear-cytoplasmic ratio, immune cell infiltration, fibrosis, and a lack of large fat droplets. CONCLUSIONS: The ability of the DL model to capture subtle pathological features beyond fibrosis suggests its potential for identifying early signs of hepatocarcinogenesis in patients with steatotic liver disease.

Conduction Velocity of Muscle Action Potential of Knee Extensor Muscle During Evoked and Voluntary Contractions After Exhaustive Leg Pedaling Exercise.

PURPOSE: Muscle fiber conduction velocity (CV) has been developed to estimate neuromuscular fatigue and measured during voluntary (VC) and electrically evoked (EC) contractions. Since CV during VC and EC reflect different physiological phenomena, the two parameters would show inconsistent changes under the conditions of neuromuscular fatigue. We investigated the time-course changes of CV during EC and VC after fatiguing exercise. METHODS: In 14 young males, maximal voluntary contraction (MVC) of knee extensor muscles, CV during electrical stimulation (CV-EC) and MVC (CV-VC) were measured before and immediately, 30 min, 60 min, 120 min, and 24 h after exhaustive leg pedaling exercise. RESULTS: CV-EC significantly increased immediately after the fatiguing exercise (p < 0.05) and had a significant negative correlation with MVC in merged data from all time-periods (r = -0.511, p < 0.001). CV-VC significantly decreased 30, 60, and 120 min after the fatiguing exercise (p < 0.05) and did not show any correlations with MVC (p > 0.05). CONCLUSION: These results suggest that CV during EC and VC exhibits different time-course changes, and that CV during EC may be appropriate to estimate the degree of neuromuscular fatigue after fatiguing pedaling exercise.