Chemistry-informed molecular graph as reaction descriptor for machine-learned retrosynthesis planning.

Infusing "chemical wisdom" should improve the data-driven approaches that rely exclusively on historical synthetic data for automatic retrosynthesis planning. For this purpose, we designed a chemistry-informed molecular graph (CIMG) to describe chemical reactions. A collection of key information that is most relevant to chemical reactions is integrated in CIMG:NMR chemical shifts as vertex features, bond dissociation energies as edge features, and solvent/catalyst information as global features. For any given compound as a target, a product CIMG is generated and exploited by a graph neural network (GNN) model to choose reaction template(s) leading to this product. A reactant CIMG is then inferred and used in two GNN models to select appropriate catalyst and solvent, respectively. Finally, a fourth GNN model compares the two CIMG descriptors to check the plausibility of the proposed reaction. A reaction vector is obtained for every molecule in training these models. The chemical wisdom of reaction propensity contained in the pretrained reaction vectors is exploited to autocategorize molecules/reactions and to accelerate Monte Carlo tree search (MCTS) for multistep retrosynthesis planning. Full synthetic routes with recommended catalysts/solvents are predicted efficiently using this CIMG-based approach.

Musculoskeletal Ultrasonography Assessment of Functional Magnetic Stimulation on the Effect of Glenohumeral Subluxation in Acute Poststroke Hemiplegic Patients.

BACKGROUND: Glenohumeral subluxation (GHS) is common in patients with acute hemiplegia caused by stroke. GHS and upper limb function are closely related. OBJECTIVE: Using musculoskeletal ultrasonography (MSUS) to objectively evaluate the efficacy of functional magnetic stimulation (FMS) in the treatment of GHS in acute hemiplegic patients after stroke. METHODS: The study used prospective case control study. Stroke patients with GHS were recruited and assigned to control group and FMS group. Control group received electrode stimulation at the supraspinatus and deltoid muscles of the hemiplegic side, while FMS group was stimulated at the same locations. Before and after treatment, the distances of the acromion-greater tuberosity (AGT), acromion-lesser tuberosity (ALT), acromiohumeral distance (AHD), supraspinatus thickness (SST), and deltoid muscle thickness (DMT) in patients' bilateral shoulder joint were measured by MSUS, respectively. Meanwhile, Fugl-Meyer Assessment (FMA) was used to evaluate the improvement of upper limb function. RESULTS: 30 patients were recruited. After FMS treatment, there was a significant decrease in the difference value between ipsilateral side and contralateral side of AGT [t = 8.595, P < 0.01], ALT [t = 11.435, P < 0.01], AHD [t = 8.375, P < 0.01], SST [t = 15.394, P < 0.01], and DMT [t = 24.935, P < 0.01], and FMA score increased [t = -13.315, P < 0.01]. Compared with control group, FMS group decreased more significantly in the difference value between ipsilateral side and contralateral side of AGT [t = 2.161, P < 0.05], ALT [t = 3.332, P < 0.01], AHD [t = 8.768, P < 0.01], SST [t = 6.244, P < 0.01], and the DMT [t = 3.238, P < 0.01], and FMA score increased more significantly in FMS group [t = 7.194, P < 0.01]. CONCLUSION: The study preliminarily shows that the MSUS can objectively and dynamically evaluate the treatment effect of GHS in hemiplegic patients. Meanwhile, compared with control group, the FMS is more effective and has fewer side effects, and the long-term effect of FMS is worth further study. This trial is registered with ChiCTR1800015352.