Yang, Deliang, Tang, Yiyi, Chan, Vivien Kin Yi, Fang, Qiwen, Chan, Sandra Sau Man, Luo, Hao, Wong, Ian Chi Kei, Ou, Huang-Tz, Chan, Esther Wai Yin, Bishai, David Makram, Chen, Yingyao, Knapp, Martin, Jit, Mark, Craig, Dawn and Li, Xue (2025). Population-Wide Depression Incidence Forecasting Comparing Autoregressive Integrated Moving Average and Vector Autoregressive Integrated Moving Average to Temporal Fusion Transformers: Longitudinal Observational Study. Journal of Medical Internet Research, 27 ,
Abstract
Accurate prediction of population-wide depression incidence is vital for effective public mental health management. However, this incidence is often influenced by socioeconomic factors, such as abrupt events or changes, including pandemics, economic crises, and social unrest, creating complex structural break scenarios in the time-series data. These structural breaks can affect the performance of forecasting methods in various ways. Therefore, understanding and comparing different models across these scenarios is essential. This study aimed to develop depression incidence forecasting models and compare the performance of autoregressive integrated moving average (ARIMA) and vector-ARIMA (VARIMA) and temporal fusion transformers (TFT) under different structural break scenarios. We developed population-wide depression incidence forecasting models and compared the performance of ARIMA and VARIMA-based methods to TFT-based methods. Using monthly depression incidence from 2002 to 2022 in Hong Kong, we applied sliding windows to segment the whole time series into 72 ten-year subsamples. The forecasting models were trained, validated, and tested on each subsample. Within each 10-year subset, the first 7 years were used for training, with the eighth year for setting hold-out validation, and the ninth and tenth years for testing. The accuracy of the testing set within each 10-year subsample was measured by symmetric mean absolute percentage error (SMAPE). We found that in subsamples without significant slope or trend change (structural break), multivariate TFT significantly outperformed univariate TFT, vector-ARIMA (VARIMA), and ARIMA, with an average SMAPE of 11.6% compared to 13.2% (P=.01) for univariate TFT, 16.4% (P=.002) for VARIMA, and 14.8% (P=.003) for ARIMA. Adjusting for the unemployment rate improved TFT performance more effectively than VARIMA. When fluctuating outbreaks happened, TFT was more robust to sharp interruptions, whereas VARIMA and ARIMA performed better when incidence surged and remained high. This study provides a comparative evaluation of TFT and ARIMA and VARIMA models for forecasting depression incidence under various structural break scenarios, offering insights into predicting disease burden during both stable and unstable periods. The findings support a decision-making framework for model selection based on the nature of disruptions and data characteristics. For public health policymaking, the results suggest that TFT may be a more suitable tool for disease burden forecasting during periods of stable burden level or when sudden temporary interruption, such as pandemics or socioeconomic variation, impacts disease occurrence.
Publication DOI: | https://doi.org/10.2196/67156 |
---|---|
Divisions: | College of Health & Life Sciences > Aston Pharmacy School |
Funding Information: | This research was supported by the Hong Kong University Grants Committee Research Impact Fund (reference number: R7007-22) and was partially supported by the Laboratory of Data Discovery for Health (D24H), funded by the AIR@InnoHK administered by the Inno |
Additional Information: | Copyright ©Deliang Yang, Yiyi Tang, Vivien Kin Yi Chan, Qiwen Fang, Sandra Sau Man Chan, Hao Luo, Ian Chi Kei Wong, Huang-Tz Ou, Esther Wai Yin Chan, David Makram Bishai, Yingyao Chen, Martin Knapp, Mark Jit, Dawn Craig, Xue Li. Originally published in the Journal of Medical Internet Research (https://www.jmir.org), 12.05.2025. This is an open-access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work, first published in the Journal of Medical Internet Research (ISSN 1438-8871), is properly cited. The complete bibliographic information, a link to the original publication on https://www.jmir.org/, as well as this copyright and license information must be included. |
Uncontrolled Keywords: | Incidence,deep learning,electronic health records,Longitudinal Studies,temporal fusion transformers,population-wide depression incidence,medical informatics,ARIMA,vector-ARIMA,Models, Statistical,depression incidence forecasting,Forecasting - methods,Hong Kong - epidemiology,machine learning,structural break scenarios,Depression - epidemiology,Humans |
Publication ISSN: | 1438-8871 |
Data Access Statement: | The data used in this study cannot be publicly shared due to third-party restrictions but are available upon request through an application to the Hong Kong Hospital Authority. The HK EHR database is managed by the Hong Kong Hospital Authority. Local academic institutions, government departments, or nongovernmental organizations can apply for access via the Hong Kong Hospital Authority Data Sharing Portal (detailed application procedures in [https://www3.ha.org.hk/data/<br/>Provision/ApplicationProcedure]). |
Last Modified: | 30 May 2025 16:01 |
Date Deposited: | 30 May 2025 11:23 |
Full Text Link: | |
Related URLs: |
https://www.jmi ... g/2025/1/e67156
(Publisher URL) |
PURE Output Type: | Article |
Published Date: | 2025-05-12 |
Published Online Date: | 2025-05-12 |
Accepted Date: | 2025-03-24 |
Authors: |
Yang, Deliang
Tang, Yiyi Chan, Vivien Kin Yi Fang, Qiwen Chan, Sandra Sau Man Luo, Hao Wong, Ian Chi Kei ( ![]() Ou, Huang-Tz Chan, Esther Wai Yin Bishai, David Makram Chen, Yingyao Knapp, Martin Jit, Mark Craig, Dawn Li, Xue |