Six-minute walking distance and desaturation-distance ratio in allogeneic stem cell transplantation.

BACKGROUND: Most patients with haematological malignancies who undergo allogeneic haematopoietic stem cell transplant (HSCT) receive chemotherapy before the transplant to control the disease. Certain chemotherapy drugs can cause lung toxicity. Conversely, in patients with chronic respiratory conditions, the 6-min walking test (6MWT) and the desaturation-distance ratio (DDR) have demonstrated prognostic significance. Our objective was to determine whether the 6MWD and DDR, assessed prior to HSCT, have a prognostic impact on survival at 24 months post-HSCT. METHODS: A prospective experimental study was conducted in consecutive patients referred for allogeneic HSCT at Hospital Clinic, Barcelona, Spain. A complete functional respiratory study, including the 6MWT and DDR, was conducted prior to admission. The area under the curve (AUC) and cut-off points were calculated. Data on patients' characteristics, HSCT details, main events, with a focus on lung complications, and survival at 24 months were analysed. RESULTS: One hundred and seventy-five patients (39% women) with mean age of 48 ± 13 years old were included. Before HSCT, forced vital capacity and forced expiratory volume in the first second were 96% ± 13% predicted and 92% ± 14% predicted, respectively; corrected diffusing capacity for carbon monoxide 79% ± 15% predicted; 6MWD was 568 ± 83 m and DDR of .27 (.20-.41). The cut-off points for 6MWD and DDR were 566 m, [.58 95% CI (.51-.64)], p = .024 and .306, [.63 95% CI (.55-.70)], p = .0005, respectively. The survival rate at 24 months was 55%. CONCLUSION: Our results showed that individuals who exhibit a 6MWD shorter than 566 ms or a decline in DDR beyond .306 experienced reduced survival rates at 24 months after HSCT.

Top-Down Machine Learning of Coarse-Grained Protein Force Fields.

Developing accurate and efficient coarse-grained representations of proteins is crucial for understanding their folding, function, and interactions over extended time scales. Our methodology involves simulating proteins with molecular dynamics and utilizing the resulting trajectories to train a neural network potential through differentiable trajectory reweighting. Remarkably, this method requires only the native conformation of proteins, eliminating the need for labeled data derived from extensive simulations or memory-intensive end-to-end differentiable simulations. Once trained, the model can be employed to run parallel molecular dynamics simulations and sample folding events for proteins both within and beyond the training distribution, showcasing its extrapolation capabilities. By applying Markov state models, native-like conformations of the simulated proteins can be predicted from the coarse-grained simulations. Owing to its theoretical transferability and ability to use solely experimental static structures as training data, we anticipate that this approach will prove advantageous for developing new protein force fields and further advancing the study of protein dynamics, folding, and interactions.