Artificial Intelligence-Based Assessment System for Evaluating Suitable Range of Heel Height

Abstract

High-heeled shoes of excessive height can severely injure shoe users. Such shoes may cause various injuries, including musculoskeletal pain, osteoarthritis, and hallux valgus. A physician can estimate an appropriate heel height limitation for an individual wearer by touching calcaneus to estimate deformation of the calcaneal varus. It would typically be impractical for a woman to seek the professional assistance of her physician when buying high-heeled shoes. A novel system was developed in this study for evaluating the maximum safe height of high-heeled shoes for female wearers. In this study, images of Achilles tendons, medial longitudinal arches, lateral longitudinal arches, and plantar pressure distributions served as the inputs in the proposed system. After the system had been trained with those images, the system could output the maximum height of high-heeled shoes for each individual wearer. In this study, two crucial methods were used for performing the evaluating system. First, Basic CNN, VGG16, and MobileNetV2 were used to evaluate images of feet. Through the experiments, the proposed artificial intelligence (AI) model achieved an accuracy of 0.88. Next, a statistics algorithm was used to modify the results obtained from the AI model. Subsequently, the error of the system declined. The mean absolute error of the proposed system which was used for evaluating the maximum height of high-heeled shoes was 1.21 cm, which is less than the typical increment for commercially available high-heeled shoes.INDEX TERMS High-heeled shoes, calcaneal varus, plantar pressure, convolutional neural network (CNN), artificial intelligence (AI)

Artificial Intelligence–Based Mobile Application for Identifying Suitable Height Range of High Heels

“…The MAE of the AI-based mobile app for evaluating the maximum height of high-heel shoes proposed in this study was 0.80 cm, which is 0.41 cm lower than the MAE (1.21 cm) of the evaluation system developed by Lee et al [7]. The reason why the MAE is slightly higher than Lee et al [7] may be that the number of photos our system took is much larger than that of [7]. If the CNN model is compared with [7] alone, the accuracy, precision, recall, and F1-score of our study are 0.98, 0.98, 0.98, and 0.98, respectively.…”

“…The reason why the MAE is slightly higher than Lee et al [7] may be that the number of photos our system took is much larger than that of [7]. If the CNN model is compared with [7] alone, the accuracy, precision, recall, and F1-score of our study are 0.98, 0.98, 0.98, and 0.98, respectively. The accuracy, precision, recall, and F1-score of [7] were 0.88, 0.86, 0.92, and 0.89, respectively.…”

“…Because the system is designed to operate on smartphones, we compared the performance of a basic, common AI model with another more suitable for smartphones (i.e., MobileNetV2). In reference to the evaluation system developed by Lee et al [7], the CNN was selected as the basic AI model, and the judgment results of the two models were evaluated.…”

“…However, the definition of an appropriate height remains unclear, and these devices cannot offer objective and quantitative evaluations for users [6]. In 2021, Lee et al proposed the first system that uses artificial intelligence (AI) to evaluate a suitable heel height for women's high-heeled shoes [7]. This system can quickly and accurately identify the most suitable high-heel height for a user and achieved an overall mean absolute error (MAE) of only 1.21 cm.…”

Application of Data Mining in Network Information Dynamic Push Software

“…Data mining, also known as knowledge discovery in databases, rose in 1989. It is the product of the integration of knowledge from multiple disciplines, including research achievements in many disciplines such as machine learning, database application technology, statistics, artificial intelligence, and so on [6].…”