Day 13. Machine learning in mental health: A systematic scoping review of methods and applications

Cite: Shatte A B R, Hutchinson D M, Teague S J. Machine learning in mental health: a scoping review of methods and applications[J]. Psychological medicine, 2019, 49(9): 1426-1448.

Title: Machine learning in mental health: A systematic scoping review of methods and applications

Abstract:
Objective: This paper aims to synthesise the literature on machine learning (ML) and big data applications for mental health, highlighting current research and applications in practice.
目的：综合有关机器学习和大数据在心理健康中的应用的文献，突出当前的研究和实际应用。
**Materials and Methods: ** Eight health and information technology research databases were searched using the terms “big data” or “machine learning” and “mental health”. Articles were assessed by two reviewers, and data were extracted on the article’s mental health application, ML technique, data type and size, and study results. Articles were then synthesised via narrative review.
材料与方法：采用大数据或机器学习、心理健康等术语检索8个健康与信息技术研究数据库。文章由两位评审员进行评估，并提取关于文章的心理健康应用、ML技术、数据类型和大小以及研究结果的数据。文章然后综合通过叙事审查。
Results: Three hundred papers focusing on the application of ML to mental health were identified. Four main application domains emerged in the literature, including: (i) detection and diagnosis; (ii) prognosis, treatment and support; (iii) public health; and, (iv) research and clinical administration. The most common mental health conditions addressed included depression, schizophrenia, and Alzheimer’s Disease. ML techniques used included support vector machines, decision trees, neural networks, latent dirichlet allocation, and clustering.
结果：共收集到300篇关于ML在心理健康领域应用的论文。文献主要出现在四个应用领域，包括：（i）检测和诊断；（ii）预测、治疗和支持；（iii）公共卫生；（iv）研究和临床管理。最常见的心理健康状况包括抑郁症、精神分裂症和老年痴呆症。ML技术包括支持向量机、决策树、神经网络、潜在狄利克雷分配和聚类。
**Discussion and Conclusion: ** Overall, the application of ML to mental health has demonstrated a range of benefits across the areas of diagnosis, treatment and support, research, and clinical administration. With the majority of studies identified focusing on the detection and diagnosis of mental health conditions, it is evident that there is significant room for the application of ML to improve other areas of psychological functioning. The challenges of using ML techniques are discussed, as well as opportunities to improve and advance the field.
讨论和结论：总的来说，应用ML进行心理健康研究在诊断、治疗和支持、研究和临床管理等方面显示了一系列的益处。由于大多数研究都集中在精神健康状况的检测和诊断上，因此很明显，ML在改善心理功能的其他方面有很大的应用空间。讨论了使用ML技术的挑战，以及改进和推进该领域的机会。

Keywords: big data, machine learning, mental health, health informatics

BACKGROUND AND SIGNIFICANCE
OBJECTIVE

Previous reviews have demonstrated ML techniques to be robust and scalable for mental health application, but no review has comprehensively mapped the clinical applications within mental health research and practice.
以往的研究表明，ML技术在心理健康领域的应用具有很强的鲁棒性和可扩展性，但目前还没有一篇综述全面描述了ML技术在心理健康研究和实践中的临床应用。

MATERIALS AND METHODS
Search strategy
Study selection
Data extraction and analysis plan
RESULTS
Overview of Article Characteristics

ML applications were also evident across a range of mental health conditions, including depression (n=88), Alzheimer’s disease and other cognitive decline (n=46), schizophrenia (n=37), stress (n=30), and suicide (n=20). The data types used to develop ML models included imaging data (n=102), survey data (n=40), mobile and wearable sensor data (n=29), and social media data (n=28), with a mean sample size of 28,754 (SD=174,426.91)
ML的应用在一系列的精神健康状况中也很显著，包括抑郁症（n=88）、阿尔茨海默病和其他认知能力下降（n=46）、精神分裂症（n=37）、压力（n=30）和自杀（n=20）。用于开发ML模型的数据类型包括成像数据（n=102）、调查数据（n=40）、移动和可穿戴传感器数据（n=29）和社交媒体数据（n=28），平均样本量为28754（SD=174426.91）。

ML Application Domains in Mental Health
Detection and Diagnosis
Prognosis, Treatment and Support

Additional benefits have been demonstrated when using ML with data from online communities, such as matching patients to suitable support communities [219] and automatic moderation of helpful comments in suicide and autism support groups [223,224].
当将ML与在线社区的数据结合使用时，例如将患者匹配到合适的支持社区 [219]，以及在自杀和自闭症支持组中自动调节有用的评论[223,224]，已经证明了其他的好处。

Public Health

Public health applications included: assessing the mental health of both specific and broader populations (e.g.[264,265]); monitoring mental health following an event or disaster (e.g.[266,267]); and creating models of risk to improve health system delivery (e.g.[268,269]) (see Table 3). Public health applications typically used social media data (n=11), electronic health records (n=6), and clinical data (e.g., diagnostic surveys and tools; n=9). Social media data was found to be a particularly useful epidemiological resource, with examples including assessments of the mental health status of over 60,000 college students in China [264] and prescription opioid misuse in an estimated sample of over 1.3 million Twitter users [265]. Social media data also enables researchers to assess the impact of an incident on population mental health (e.g., stress levels of college students after experiencing gun violence [270]), and to track public response to disaster situations to inform the allocation of support resources [266,267,269]. ML applied to electronic health records was demonstrated to predict suicide risk with an accuracy similar to clinician assessment [268,271], as well as predict dementia and its risk factors with high accuracy [272]. Research has also investigated the use of ML with clinical data to improve variable selection in epidemiological data analysis [273], and to better understand the relationship between complex risk factors for mental health conditions such as depression [274].
公共卫生应用包括：评估特定人群和更广泛人群的心理健康(e.g.[264,265])；监测事件或灾难后的心理健康(e.g.[266,267]); 建立风险模型以改善卫生系统交付(e.g.[268,269])（见表3）。公共卫生的应用通常使用社交媒体数据（n=11）、电子健康记录（n=6）和临床数据（例如，诊断调查和工具；n=9）。社交媒体数据被认为是一种特别有用的流行病学资源，其中包括对中国6万多名大学生的心理健康状况的评估 [264]，以及对130多万推特用户的处方类阿片滥用的估计样本[265]。社交媒体数据还使研究人员能够评估特定事件对群众心理健康的影响（例如，大学生遭遇枪支暴力事件后的压力水平[270]），并跟踪公众对灾难情况的反应，以调整有效资源的分配[266,267,269]。将ML应用于电子健康档案，可以预测自杀风险，其准确度与临床医生的评估相似[268,271]，也可以预测痴呆症及其危险因素，具有较高的准确度[272]。研究还调查了使用ML结合临床数据改进变量选择的流行病学数据分析[273]，并更好地了解抑郁症等精神健康状况复杂危险因素之间的关系 [274]。

Overall, ML appears to be a promising tool for public health. Social media data and electronic health records are enabling researchers to monitor the wellbeing of large groups of people in a cost-efficient manner. Social media data in particular is providing an ecologically valid assessment of mental health in the population in real-time, enabling assessment of groups that have typically been challenging to monitor through traditional research methods (e.g., opioid misuse [265]).
总的来说，ML是一个有前途的研究公共卫生的工具。社交媒体数据和电子健康记录使研究人员能够以一种成本效益高的方式监测大量人群的健康状况。尤其是社交媒体数据提供了对群众心理健康的生态有效的实时评估，使人们能够评估那些通常难以通过传统研究方法进行监测的群体（如阿片类药物滥用[265]）。

Research and Clinical Administration
DISCUSSION
CONCLUSION