Automated entry of paper-based patient-reported outcomes: Applying deep learning to the Japanese orthopaedic association back pain evaluation questionnaire

• Published in: Computers in biology and medicine Vol. 172; p. 108197
• Main Authors: Kita, Kosuke, Fujimori, Takahito, Suzuki, Yuki, Kaito, Takashi, Takenaka, Shota, Kanie, Yuya, Furuya, Masayuki, Wataya, Tomohiro, Nishigaki, Daiki, Sato, Junya, Tomiyama, Noriyuki, Okada, Seiji, Kido, Shoji
• Format: Journal Article
• Language: English
• Published: United States Elsevier Ltd 01.04.2024; Elsevier Limited
• Subjects: Accuracy; Artificial intelligence; Automation; Back Pain; Clinical outcomes; Community centers; Convolutional neural network; Correlation coefficient; Correlation coefficients; Data collection; Data entry; Decision making; Deep Learning; Forms (paper); Health care facilities; HR-PRO; Humans; Japan; JOABPEQ; Low back pain; Low Back Pain - diagnosis; Low Back Pain - therapy; Machine learning; Mathematical models; Medical equipment; Model accuracy; Older people; Orthopedics; Pain; Patients; Performance evaluation; Probability; Quality assessment; Quality of Life; Questionnaire; Questionnaires; Questions; Statistical analysis; Surveys and Questionnaires; Japan; Deep learning; Back pain; Questionnaire; Convolutional neural network; JOABPEQ; HR-PRO; Artificial intelligence
• ISSN: 0010-4825; 1879-0534
• DOI: 10.1016/j.compbiomed.2024.108197

Health-related patient-reported outcomes (HR-PROs) are crucial for assessing the quality of life among individuals experiencing low back pain. However, manual data entry from paper forms, while convenient for patients, imposes a considerable tallying burden on collectors. In this study, we developed a deep learning (DL) model capable of automatically reading these paper forms. We employed the Japanese Orthopaedic Association Back Pain Evaluation Questionnaire, a globally recognized assessment tool for low back pain. The questionnaire comprised 25 low back pain-related multiple-choice questions and three pain-related visual analog scales (VASs). We collected 1305 forms from an academic medical center as the training set, and 483 forms from a community medical center as the test set. The performance of our DL model for multiple-choice questions was evaluated using accuracy as a categorical classification task. The performance for VASs was evaluated using the correlation coefficient and absolute error as regression tasks. In external validation, the mean accuracy of the categorical questions was 0.997. When outputs for categorical questions with low probability (threshold: 0.9996) were excluded, the accuracy reached 1.000 for the remaining 65 % of questions. Regarding the VASs, the average of the correlation coefficients was 0.989, with the mean absolute error being 0.25. Our DL model demonstrated remarkable accuracy and correlation coefficients when automatic reading paper-based HR-PROs during external validation. •A deep learning model for reading patient reported outcomes.•We propose a deep learning model capable of processing an entire form.•The proposed model accurately extracting information from paper-based patient reported outcomes.