Purpose/Significance To explore the important factors affecting the connotation quality of inpatient medical records, and to provide model prediction and improve the connotation quality of inpatient medical records. Method/Process A total of 590 inpatient medical records monitored by quality control in Shanghai First People’s Hospital from June to November 2022 are collected. The influencing factors are initially screened by single factor analysis, and a multi-layer perceptron neural network prediction model for the connotation quality of inpatient medical records is constructed. Result/Conclusion The area under the curve (AUC) of the prediction model is 0.940,95% CI is 0.928~0.951, the sensitivity is 93.73%, and the specificity is 78.22%. The top three independent factors affecting the rating of a case as grade A are concentrated in the surgical safety checklist, the analysis of the first director’s ward round, and the surgical nursing record. The multi-layer perceptron neural network connotation quality prediction model has good prediction efficiency, which provides theoretical references for the connotation quality management of inpatient medical records.