AI Enhanced Tai Chi Rehabilitation for Substance Use Disorder with Clinical Evidence and Predictive Modeling for Relapse Prevention
DOI:
https://doi.org/10.70062/jmih.v1i2.170Keywords:
Substance use disorder, Tai Chi, artificial intelligence, rehabilitation, relapse prediction, human activity recognition, psychometrics, deep learningAbstract
Substance use disorder (SUD) remains a significant global health issue, with relapse rates exceeding 60% in some compulsory rehabilitation centers despite structured interventions. Recent research suggests that mind–body exercises, like Tai Chi, can lessen cravings and enhance psychological well-being, though their current use is often standardized, subjective, and not personalized. This study aimed to evaluate the clinical benefits of Tai Chi within a compulsory rehabilitation setting and to develop an AI-enhanced framework for individualized relapse prevention. A randomized controlled trial (RCT) involving 168 participants diagnosed with SUD compared a Tai Chi intervention group with a standard physical education control group. Psychometric outcomes were measured using validated scales, while a conceptual AI framework using a CNN-LSTM architecture was simulated with multimodal inputs that combined psychometric indicators and motion-based features. The RCT showed significant reductions in craving, depression, and anxiety, along with improved self-control in the Tai Chi group compared to controls. Mediation analysis indicated that psychological symptoms partially explained the link between craving and relapse risk. The AI simulation achieved 82% accuracy and an area under the ROC curve of 0.85, with craving and depression identified as key predictors. These results provide initial evidence that integrating Tai Chi with AI-driven monitoring can transform exercise-based rehabilitation into a closed-loop, adaptive system capable of delivering personalized feedback and early relapse warnings. This combined approach has the potential to enhance the scalability, accuracy, and effectiveness of institutional rehab programs for SUD.
References
[1] Z. Tang, Z. Zhu, X. Zhang, and Y. He, “The effects of Tai Chi and Baduanjin activities on physical interventions with substance use disorders: A systematic review and meta-analysis,” Hong Kong J. Occup. Ther., vol. 35, pp. 47–56, May 2022, doi: 10.1177/15691861221090551.
[2] F. Liu, J. Cui, X. Liu, K. W. Chen, X. Chen, and R. Li, “The effect of Tai Chi and Qigong exercise on depression and anxiety of individuals with substance use disorders: A systematic review and meta-analysis,” BMC Complement. Med. Ther., vol. 20, no. 1, p. 182, May 2020, doi: 10.1186/s12906-020-02967-8.
[3] J. Cui et al., “The impact of Qigong and Tai Chi exercise on drug addiction: A systematic review and meta-analysis,” Front. Psychiatry, vol. 13, p. 826187, Mar. 2022, doi: 10.3389/fpsyt.2022.826187.
[4] M. Wang et al., “A randomized controlled trial evaluating the effect of Tai Chi on the drug craving in women,” Int. J. Ment. Health Addict., Sep. 2022, doi: 10.1007/s11469-022-00917-8.
[5] G. Perna, M. Grassi, D. Caldirola, and C. B. Nemeroff, “The revolution of personalized psychiatry: Will technology make it happen sooner?,” Psychol. Med., vol. 48, no. 5, pp. 705–713, Oct. 2017, doi: 10.1017/S0033291717002859.
[6] D. A. Adler et al., “Predicting early warning signs of psychotic relapse from passive sensing data: An approach using encoder-decoder neural networks,” JMIR Mhealth Uhealth, vol. 8, no. 8, p. e19962, Aug. 2020, doi: 10.2196/19962.
[7] N. Koutsouleris, T. U. Hauser, V. Skvortsova, and M. De Choudhury, “From promise to practice: Towards the realisation of AI-informed mental health care,” Lancet Digit. Health, vol. 4, no. 10, pp. e727–e739, Oct. 2022, doi: 10.1016/S2589-7500(22)00153-4.
[8] S. Mekruksavanich, A. Jitpattanakul, P. Youplao, and P. Yupapin, “Enhanced hand-oriented activity recognition based on smartwatch sensor data using LSTMs,” Symmetry, vol. 12, no. 9, p. 1570, Sep. 2020, doi: 10.3390/sym12091570.
[9] H. Wang et al., “Wearable sensor‐based human activity recognition using hybrid deep learning techniques,” Secur. Commun. Netw., vol. 2020, pp. 1–11, Jul. 2020, doi: 10.1155/2020/2132138.
[10] N. Moukaddam, A. Sano, R. Salas, Z. Hammal, and A. Sabharwal, “Turning data into better mental health: Past, present, and future,” Front. Digit. Health, vol. 4, p. 916810, Aug. 2022, doi: 10.3389/fdgth.2022.916810.
[11] A. R. Zhang et al., “Cocaine use prediction with tensor-based machine learning on multimodal MRI connectome data,” Neural Comput., vol. 35, no. 12, pp. 2609–2633, Dec. 2023, doi: 10.1162/neco_a_01623.
[12] X. Tong et al., “Optimizing antidepressant efficacy: Generalizable multimodal neuroimaging biomarkers for prediction of treatment response,” medRxiv, Apr. 2024, doi: 10.1101/2024.04.11.24305583.
[13] E. Mikołajewska and J. Masiak, “Deep learning approaches to natural language processing for digital twins of patients in psychiatry and neurological rehabilitation,” Electronics, vol. 14, no. 10, p. 2024, May 2025, doi: 10.3390/electronics14102024.
[14] Y. Daineko and A. Seitnur, “Artificial intelligence for predictive health analytics: Challenges, methods, and immersive system integration,” Procedia Comput. Sci., vol. 218, pp. 1709–1716, Jan. 2025, doi: 10.1016/j.procs.2025.07.232.
[15] C. Kokkotis et al., “Artificial intelligence as assessment tool in occupational therapy: A scoping review,” BioMedInformatics, vol. 5, no. 2, p. 22, Apr. 2025, doi: 10.3390/biomedinformatics5020022.
