Machine Learning and Natural Language Processing in Mental Health: Systematic Review.

BACKGROUND: Machine learning systems are part of the field of artificial intelligence that automatically learn models from data to make better decisions. Natural language processing (NLP), by using corpora and learning approaches, provides good performance in statistical tasks, such as text classification or sentiment mining. OBJECTIVE: The primary aim of this systematic review was to summarize and characterize, in methodological and technical terms, studies that used machine learning and NLP techniques for mental health. The secondary aim was to consider the potential use of these methods in mental health clinical practice. METHODS: This systematic review follows the PRISMA (Preferred Reporting Items for Systematic Review and Meta-analysis) guidelines and is registered with PROSPERO (Prospective Register of Systematic Reviews; number CRD42019107376). The search was conducted using 4 medical databases (PubMed, Scopus, ScienceDirect, and PsycINFO) with the following keywords: machine learning, data mining, psychiatry, mental health, and mental disorder. The exclusion criteria were as follows: languages other than English, anonymization process, case studies, conference papers, and reviews. No limitations on publication dates were imposed. RESULTS: A total of 327 articles were identified, of which 269 (82.3%) were excluded and 58 (17.7%) were included in the review. The results were organized through a qualitative perspective. Although studies had heterogeneous topics and methods, some themes emerged. Population studies could be grouped into 3 categories: patients included in medical databases, patients who came to the emergency room, and social media users. The main objectives were to extract symptoms, classify severity of illness, compare therapy effectiveness, provide psychopathological clues, and challenge the current nosography. Medical records and social media were the 2 major data sources. With regard to the methods used, preprocessing used the standard methods of NLP and unique identifier extraction dedicated to medical texts. Efficient classifiers were preferred rather than transparent functioning classifiers. Python was the most frequently used platform. CONCLUSIONS: Machine learning and NLP models have been highly topical issues in medicine in recent years and may be considered a new paradigm in medical research. However, these processes tend to confirm clinical hypotheses rather than developing entirely new information, and only one major category of the population (ie, social media users) is an imprecise cohort. Moreover, some language-specific features can improve the performance of NLP methods, and their extension to other languages should be more closely investigated. However, machine learning and NLP techniques provide useful information from unexplored data (ie, patients' daily habits that are usually inaccessible to care providers). Before considering It as an additional tool of mental health care, ethical issues remain and should be discussed in a timely manner. Machine learning and NLP methods may offer multiple perspectives in mental health research but should also be considered as tools to support clinical practice.

Assessment of Selective and Universal Screening for Suicide Risk in a Pediatric Emergency Department.

Importance: According to National Patient Safety Goal 15.01.01, all individuals being treated or evaluated for behavioral health conditions as their primary reason for care in hospitals and behavioral health care organizations accredited by The Joint Commission should be screened for suicide risk using a validated tool. Existing suicide risk screens have minimal or no high-quality evidence of association with future suicide-related outcomes. Objective: To test the association between results of the Ask Suicide-Screening Questions (ASQ) instrument in a pediatric emergency department (ED), implemented through selective and universal screening approaches, and subsequent suicide-related outcomes. Design Setting and Participants: In this retrospective cohort study at an urban pediatric ED in the United States, the ASQ was administered to youths aged 8 to 18 years with behavioral and psychiatric presenting problems from March 18, 2013, to December 31, 2016 (selective condition), and then to youths aged 10 to 18 years with medical presenting problems (in addition to those aged 8-18 years with behavioral and psychiatric presenting problems) from January 1, 2017, to December 31, 2018 (universal condition). Exposure: Positive ASQ screen at baseline ED visit. Main Outcomes and Measures: The main outcomes were subsequent ED visits with suicide-related presenting problems (ie, ideation or attempts) based on electronic health records and death by suicide identified through state medical examiner records. Association with suicide-related outcomes was calculated over the entire study period using survival analyses and at 3-month follow-up for both conditions using relative risk. Results: The complete sample was 15003 youths (7044 47.0%] male; 10209 [68.0%] black; mean [SD] age, 14.5 [3.1] years at baseline). The follow-up for the selective condition was a mean (SD) of 1133.7 (433.3) days; for the universal condition, it was 366.2 (209.2) days. In the selective condition, there were 275 suicide-related ED visits and 3 deaths by suicide. In the universal condition, there were 118 suicide-related ED visits and no deaths during the follow-up period. Adjusting for demographic characteristics and baseline presenting problem, positive ASQ screens were associated with greater risk of suicide-related outcomes among both the universal sample (hazard ratio, 6.8 [95% CI, 4.2-11.1]) and the selective sample (hazard ratio, 4.8 [95% CI, 3.5-6.5]). Conclusions and Relevance: Positive results of both selective and universal screening for suicide risk in pediatric EDs appear to be associated with subsequent suicidal behavior. Screening may be a particularly effective way to detect suicide risk among those who did not present with ideation or attempt. Future studies should examine the impact of screening in combination with other policies and procedures aimed at reducing suicide risk.Appeared originally in JAMA Netw Open 2019; 2:e1914070.

A Brief Text-Messaging Intervention for Suicidal Youths After Emergency Department Discharge.

Research has shown that the months after hospital discharge following treatment for suicidal thoughts or behaviors is a high-risk period for suicide. Moreover, the needs of a subset of youths at increased risk for suicide are not being met by community mental health providers, resulting in frequent emergency department (ED) visits. While undertaking a quality improvement effort, the authors' health care system piloted caring text messages to support youths discharged from the hospital after screening positive for suicide risk in the ED. The text-messaging intervention was feasible and acceptable, and youths reported that the messages helped reduce their suicidal thoughts and behaviors postdischarge. The results of this text-messaging intervention prompted the Maryland Department of Health to offer a similar intervention to all Marylanders.

Ineligibility for and Refusal to Participate in Randomized Controlled Trials That Have Studied Impact on Suicide-Related Outcomes in the United States: A Meta-Analysis.

Objective: Ineligibility for and refusal to participate in randomized controlled trials (RCTs) can potentially lead to unrepresentative study samples and limited generalizability of findings. We examined the rates of exclusion and refusal in RCTs that have studied impact on suicide-related outcomes in the US.Data Sources: PubMed, the Cochrane Library, the Campbell Collaboration Library of Systematic Reviews, CINAHL, PsycINFO, and Education Resources Information Center were searched from January 1990 to May 2020 using the terms (suicide prevention) AND (clinical trial).Study Selection: Of 8,403 studies retrieved, 36 RCTs assessing effectiveness on suicide-related outcomes in youth (≤ 25 years old) conducted in the US were included.Data Extraction: Study-level data were extracted by 2 independent investigators for a random-effects meta-analysis and meta-regression.Results: The study participants (N = 13,264) had a mean (SD) age of 14.87 (1.58) years and were 50% male, 23% African American, and 24% Hispanic. The exclusion rate was 36.4%, while the refusal rate was 25.5%. The exclusion rate was significantly higher in the studies excluding individuals not exceeding specified cutoff points of suicide screening tools (51.2%; adjusted linear coefficient [ß] = 1.30, standard error [SE] = 0.15; P = .041) and individuals not meeting the age or school grade criterion (45.9%; ß = 1.37, SE = 0.13; P = .005).Conclusions: The rates of exclusion and refusal in youth prevention interventions studying impact on suicide-related outcomes were not as high compared to the rates found in other mental and behavioral interventions. While there was strong racial/ethnic group representation in RCTs examining youth suicide-related outcomes, suicide severity and age limited eligibility.

Suicide Risk Screening in Pediatric Outpatient Neurodevelopmental Disabilities Clinics.

OBJECTIVE: The purpose of this study was to describe the implementation of universal suicide risk screening in pediatric neurodevelopmental disabilities (NDD) medical clinics, analyze demographic and clinical characteristics of eligible patients, describe outcomes of positive screenings, and describe factors that influenced participation in screenings. METHODS: A suicide risk screening protocol was developed and implemented for medical clinic patients aged 8 to 18 years. Registered nurses screened patients using the "Ask Suicide-Screening Questions" tool during triage. Positive screenings were referred for further assessment and mental health management. Demographics and clinical data were extracted from medical records using retrospective chart reviews. RESULTS: During the 6-month study period, 2961 individual patients presented for 5260 screening eligible patient visits. In total, 3854 (73.3%) screenings were completed with 261 (6.8%) positive screenings noted. Screenings were declined in 1406 (26.7%) visits. Parents of children with cognitive impairments were more likely to decline screening. Clinics serving children with autism spectrum disorder had higher rates of positive screenings compared with all other clinic attendees. Seventy-two of 187 children (38.5%) with positive screenings were identified and referred to outpatient mental health referrals. Seven (2.5%) of these children required acute psychiatric treatment. CONCLUSION: Routine screening, identification of increased suicide risk, and referral to mental health care among children with NDD are feasible. It remains unclear whether variation in rates among youth with and without NDD may indicate true differences in suicide risk or cognitive impairments or reflect psychiatric comorbidities. High rates of declined participation may have influenced identification of children with NDD and suicide risk. Preliminary findings identified groups of children with NDD at heightened risk for suicidal ideation and behavior. Further research is needed to assess the validity of suicide risk screening tools in children with neurodevelopmental disorders.

Lowering the Age Limit in Suicide Risk Screening: Clinical Differences and Screening Form Predictive Ability.

Our research provides preliminary evidence that suicide risk screening is warranted in patients as young as 8-9 years old presenting to the emergency department (ED) with behavioral and mental health symptoms. The goal of this retrospective cohort study (N = 2,466 unique patient visits) was to assess the value of suicide risk screening in children younger than 10 years old who present to the ED with behavioral and mental health concerns. The Johns Hopkins Hospital pediatric ED began screening with the Ask Suicide-Screening Questions (ASQ) for patients 8-21 years old who presented with a behavioral or mental health concern in March 2013 as ED standard of care. We examined the demographic and clinical differences between younger (8-9 years old; n = 270) and older (10-21 years old; n = 2,196) youths who were screened for suicide risk with the ASQ (from March 13, 2013 through December 31, 2016). In summary, 36% of 8- and 9-year-old patients who came to the ED for behavioral and mental health concerns screened positive for suicide risk on the ASQ. The younger patients who screened positive were more likely to present with externalizing symptoms and hallucinations and less likely to present with suicidal ideation or an attempt than their older counterparts. Importantly, 71.1% of 8- to 9-year-old patients who screened positive did not present to the ED for suicidal ideation or attempt vs 50.1% (614/1,226) of patients older than age 10 years.

Screening for Suicide Risk Among Youths With a Psychotic Disorder in a Pediatric Emergency Department.

National Patient Safety Goal 15.01.01 requires all Joint Commission-accredited organizations to screen patients treated for behavioral health conditions for suicide risk. However, little is known about the ability of screening tools to identify suicide risk among youths with psychotic disorders. As part of this quality improvement initiative, youths in a pediatric emergency department with psychotic disorder diagnoses (N=87) were screened with the Ask Suicide-Screening Questions. Almost half (48%, N=42) screened positive. Most positive screens (62%, N=26) were not detected through treatment as usual, suggesting that systematic screening improves the detection of suicide risk among youths with psychotic disorders.

Suicide Screening in Sturge-Weber Syndrome: An Important Issue in Need of Further Study.

BACKGROUND: Sturge-Weber syndrome is a neurocutaneous disorder associated with epilepsy, glaucoma, cognitive impairments, and a port-wine birthmark. Although individuals with Sturge-Weber syndrome are vulnerable to known risk factors for suicide, including chronic illness and physical differences (port-wine birthmark), frequency of suicidal ideation and attempts, and the clinical factors associated with suicide risk, in patients with Sturge-Weber syndrome is unknown. METHODS: As a part of routine hospital practice, all outpatients aged eight years and older underwent suicide risk screening during nursing triage using a standardized suicide screening tool. Suicide risk screening results, demographic variables, and medical history (as available) for patients with Sturge-Weber syndrome (N = 34; median age = 15.5; range = 8 to 47 years, 44% male) and other neurological conditions seen at the same institution (N = 369; median age = 14; range = 8 to 78 years, 66% male) were used for retrospective within- and between-group analysis. RESULTS: In the combined sample of Sturge-Weber syndrome and neurologically involved patients, a positive suicide risk screen was related to Sturge-Weber syndrome diagnosis (P = 0.043); analysis by sex showed increased risk of Sturge-Weber syndrome diagnosis in males (P = 0.008), but not in females. Within the Sturge-Weber syndrome group, use of a selective serotonin reuptake inhibitor (P = 0.019) was related to a positive risk screen. CONCLUSION: People with Sturge-Weber syndrome may be at greater risk of suicidal thoughts or behaviors than those with other neurological conditions. Further study of suicide risk in patients with Sturge-Weber syndrome is needed.

Assessment of Selective and Universal Screening for Suicide Risk in a Pediatric Emergency Department.

Importance: According to National Patient Safety Goal 15.01.01, all individuals being treated or evaluated for behavioral health conditions as their primary reason for care in hospitals and behavioral health care organizations accredited by The Joint Commission should be screened for suicide risk using a validated tool. Existing suicide risk screens have minimal or no high-quality evidence of association with future suicide-related outcomes. Objective: To test the association between results of the Ask Suicide-Screening Questions (ASQ) instrument in a pediatric emergency department (ED), implemented through selective and universal screening approaches, and subsequent suicide-related outcomes. Design, Setting, and Participants: In this retrospective cohort study at an urban pediatric ED in the United States, the ASQ was administered to youths aged 8 to 18 years with behavioral and psychiatric presenting problems from March 18, 2013, to December 31, 2016 (selective condition), and then to youths aged 10 to 18 years with medical presenting problems (in addition to those aged 8-18 years with behavioral and psychiatric presenting problems) from January 1, 2017, to December 31, 2018 (universal condition). Exposure: Positive ASQ screen at baseline ED visit. Main Outcomes and Measures: The main outcomes were subsequent ED visits with suicide-related presenting problems (ie, ideation or attempts) based on electronic health records and death by suicide identified through state medical examiner records. Association with suicide-related outcomes was calculated over the entire study period using survival analyses and at 3-month follow-up for both conditions using relative risk. Results: The complete sample was 15 003 youths (7044 [47.0%] male; 10 209 [68.0%] black; mean [SD] age, 14.5 [3.1] years at baseline). The follow-up for the selective condition was a mean (SD) of 1133.7 (433.3) days; for the universal condition, it was 366.2 (209.2) days. In the selective condition, there were 275 suicide-related ED visits and 3 deaths by suicide. In the universal condition, there were 118 suicide-related ED visits and no deaths during the follow-up period. Adjusting for demographic characteristics and baseline presenting problem, positive ASQ screens were associated with greater risk of suicide-related outcomes among both the universal sample (hazard ratio, 6.8 [95% CI, 4.2-11.1]) and the selective sample (hazard ratio, 4.8 [95% CI, 3.5-6.5]). Conclusions and Relevance: Positive results of both selective and universal screening for suicide risk in pediatric EDs appear to be associated with subsequent suicidal behavior. Screening may be a particularly effective way to detect suicide risk among those who did not present with ideation or attempt. Future studies should examine the impact of screening in combination with other policies and procedures aimed at reducing suicide risk.