Beyond black-box models: explainable AI for embryo ploidy prediction and patient-centric consultation.

PURPOSE: To determine if an explainable artificial intelligence (XAI) model enhances the accuracy and transparency of predicting embryo ploidy status based on embryonic characteristics and clinical data. METHODS: This retrospective study utilized a dataset of 1908 blastocyst embryos. The dataset includes ploidy status, morphokinetic features, morphology grades, and 11 clinical variables. Six machine learning (ML) models including Random Forest (RF), Linear Discriminant Analysis (LDA), Logistic Regression (LR), Support Vector Machine (SVM), AdaBoost (ADA), and Light Gradient-Boosting Machine (LGBM) were trained to predict ploidy status probabilities across three distinct datasets: high-grade embryos (HGE, n = 1107), low-grade embryos (LGE, n = 364), and all-grade embryos (AGE, n = 1471). The model's performance was interpreted using XAI, including SHapley Additive exPlanations (SHAP) and Local Interpretable Model-agnostic Explanations (LIME) techniques. RESULTS: The mean maternal age was 38.5 ± 3.85 years. The Random Forest (RF) model exhibited superior performance compared to the other five ML models, achieving an accuracy of 0.749 and an AUC of 0.808 for AGE. In the external test set, the RF model achieved an accuracy of 0.714 and an AUC of 0.750 (95% CI, 0.702-0.796). SHAP's feature impact analysis highlighted that maternal age, paternal age, time to blastocyst (tB), and day 5 morphology grade significantly impacted the predictive model. In addition, LIME offered specific case-ploidy prediction probabilities, revealing the model's assigned values for each variable within a finite range. CONCLUSION: The model highlights the potential of using XAI algorithms to enhance ploidy prediction, optimize embryo selection as patient-centric consultation, and provides reliability and transparent insights into the decision-making process.

ETF-QO Mutants Uncoupled Fatty Acid ß-Oxidation and Mitochondrial Bioenergetics Leading to Lipid Pathology.

The electron-transfer flavoprotein dehydrogenase gene (ETFDH) that encodes the ETF-ubiquinone oxidoreductase (ETF-QO) has been reported to be the major cause of multiple acyl-CoA dehydrogenase deficiency (MADD). ETF-QO is an electron carrier that mainly functions in mitochondrial fatty acid ß-oxidation and the delivery of electrons to the ubiquinone pool in the mitochondrial respiratory chain. A high frequency of c.250G>A has been found in Taiwanese patients with late-onset MADD. We postulated that the ETFDH c.250G>A mutation may concomitantly impair fatty acid ß-oxidation and mitochondrial function. Using MADD patient-derived lymphoblastoid cells and specifically overexpressed ETFDH c.92C>T, c.250G>A, or coexisted c.92C>T and c.250G>A (c.92C>T + c.250G>A) mutated lymphoblastoid cells, we addressed the genotype-phenotype relationship of ETFDH variation in the pathogenesis of MADD. The decreased adenosine triphosphate synthesis, dissipated mitochondrial membrane potentials, reduced mitochondrial bioenergetics, and increased neutral lipid droplets and lipid peroxides were found in the MADD patient-derived lymphoblastoid cells. Riboflavin and/or coenzyme Q10 supplementation rescued cells from lipid droplet accumulation. All three mutant types, c.92C>T, c.250G>A, or c.92C>T + c.250G>A, had increased lipid droplet accumulation after treatment with palmitic acid. These results help to clarify the molecular pathogenesis of MADD as a result of the high frequency of the ETFDH c.250G>A and c.92C>T mutations.

Coenzyme Q10 serves to couple mitochondrial oxidative phosphorylation and fatty acid ß-oxidation, and attenuates NLRP3 inflammasome activation.

Multiple acyl-CoA dehydrogenase deficiency (MADD), an autosomal recessive metabolic disorder of fatty acid metabolism, is mostly caused by mutations in the ETFA, ETFB or ETFDH genes that result in dysfunctions in electron transfer flavoprotein (ETF) or electron transfer flavoprotein-ubiquinone dehydrogenase (ETFDH). In ß-oxidation, fatty acids are processed to generate acyl-CoA, which is oxidised by flavin adenine dinucleotide and transfers an electron to ETF and, through ETFDH, to mitochondrial respiratory complex III to trigger ATP synthesis. Coenzyme Q10 (CoQ10) is believed to be a potential treatment that produces symptom relief in some MADD patients. CoQ10 acts as a key regulator linking ETFDH and mitochondrial respiratory complex III. Our aim is to investigate the effectiveness of CoQ10 in serving in the ETF/ETFDH system to improve mitochondrial function and to reduce lipotoxicity. In this study, we used lymphoblastoid cells with an ETFDH mutation from MADD patients. ETFDH dysfunction caused insufficient ß-oxidation, leading to increasing lipid droplet and lipid peroxide accumulation. In contrast, supplementation with CoQ10 significantly recovered mitochondrial function and concurrently decreased the generation of reactive oxygen species and lipid peroxides, inhibited the accumulation of lipid droplets and the formation of the NOD-like receptor family pyrin domain-containing three (NLRP3) inflammasome, and reduced interleukin-1ß release and cell death. These results clarify the causal role of CoQ10 in coupling the electron transport chain with ß-oxidation, which may promote the development of CoQ10-directed therapies for MADD patients.

Cerebral microvascular damage occurs early after hypoxia-ischemia via nNOS activation in the neonatal brain.

Microvascular injury early after hypoxic ischemia (HI) may contribute to neonatal brain damage. N-methyl-D-aspartate receptor overstimulation activates neuronal nitric oxide synthases (nNOS). We hypothesized that microvascular damage occurs early post-HI via nNOS activation and contributes to brain injury. Postpartum day-7 rat pups were treated with 7-nitroindazole (7-NI) or aminoguanidine (AG) before or after HI. Electron microscopy was performed to measure neuronal and endothelial cell damage. There were vascular lumen narrowing at 1 hour, pyknotic neurons at 3 hours, and extensive neuronal damage and loss of vessels at 24 hours post HI. Early after reoxygenation, there were neurons with heterochromatic chromatin and endothelial cells with enlarged nuclei occluding the lumen. There was also increased 3-nitrotyrosin in the microvessels and decreased cerebral blood perfusion. 7-NI and AG treatment before hypoxia provided complete and partial neuroprotection, respectively. Early post-reoxygenation, the AG group showed significantly increased microvascular nitrosative stress, microvascular interruptions, swollen nuclei that narrowed the vascular lumen, and decreased cerebral perfusion. The 7-NI group showed significantly decreased microvascular nitrosative stress, patent vascular lumen, and increased cerebral perfusion. Our results indicate that microvascular damage occurs early and progressively post HI. Neuronal nitric oxide synthases activation contributes to microvascular damage and decreased cerebral perfusion early after reoxygenation and worsens brain damage.