Predicting time to relapse in patients with schizophrenia according to patients' relapse history: a historical cohort study using real-world data in Sweden

Accuracy of machine learning methods in predicting prognosis of patients with psychotic spectrum disorders: a systematic review

OBJECTIVES

We aimed to examine the predictive accuracy of functioning, relapse or remission among patients with psychotic disorders, using machine learning methods. We also identified specific features that were associated with these clinical outcomes.

DESIGN

The methodology of this review was guided by the Cochrane Handbook for Systematic Reviews of Diagnostic Test Accuracy.

DATA SOURCES

CINAHL, EMBASE, PubMed, PsycINFO, Scopus and ScienceDirect were searched for relevant articles from database inception until 21 November 2024.

ELIGIBILITY CRITERIA

Studies were included if they involved the use of machine learning methods to predict functioning, relapse and/or remission among individuals with psychotic spectrum disorders.

DATA EXTRACTION AND SYNTHESIS

Two independent reviewers screened the records from the database search. Risk of bias was evaluated using the Quality Assessment of Diagnostic Accuracy Studies tool from Cochrane. Synthesised findings were presented in tables.

RESULTS

23 studies were included in the review, which were mostly conducted in the west (91%). Predictive summary area under the curve values for functioning, relapse and remission were 0.63-0.92 (poor to outstanding), 0.45-0.95 (poor to outstanding), 0.70-0.79 (acceptable), respectively. Logistic regression and random forest were the best performing algorithms. Factors influencing outcomes included demographic (age, ethnicity), illness (duration of untreated illness, types of symptoms), functioning (baseline functioning, interpersonal relationships and activity engagement), treatment variables (use of higher doses of antipsychotics, electroconvulsive therapy), data from passive sensor (call log, distance travelled, time spent in certain locations) and online activities (time of use, use of certain words, changes in search frequencies and length of queries).

CONCLUSION

Machine learning methods show promise in the prediction of prognosis (specifically functioning, relapse and remission) of mental disorders based on relevant collected variables. Future machine learning studies may want to focus on the inclusion of a broader swathe of variables including ecological momentary assessments, with a greater amount of good quality big data covering longer longitudinal illness courses and coupled with external validation of study findings.

PROSPERO REGISTRATION NUMBER

The review was registered on PROSPERO, ID: CRD42023441108.

Predictive utility of artificial intelligence on schizophrenia treatment outcomes: A systematic review and meta-analysis

Identifying optimal treatment approaches for schizophrenia is challenging due to varying symptomatology and treatment responses. Artificial intelligence (AI) shows promise in predicting outcomes, prompting this systematic review and meta-analysis to evaluate various AI models' predictive utilities in schizophrenia treatment. A systematic search was conducted, and the risk of bias was evaluated. The pooled sensitivity, specificity, and diagnostic odds ratio with 95 % confidence intervals between AI models and the reference standard for response to treatment were assessed. Diagnostic accuracy measures were calculated, and subgroup analysis was performed based on the input data of AI models. Out of the 21 included studies, AI models achieved a pooled sensitivity of 70 % and specificity of 76 % in predicting schizophrenia treatment response with substantial predictive capacity and a near-to-high level of test accuracy. Subgroup analysis revealed EEG-based models to have the highest sensitivity (89 %) and specificity (94 %), followed by imaging-based models (76 % and 80 %, respectively). However, significant heterogeneity was observed across studies in treatment response definitions, participant characteristics, and therapeutic interventions. Despite methodological variations and small sample sizes in some modalities, this study underscores AI's predictive utility in schizophrenia treatment, offering insights for tailored approaches, improving adherence, and reducing relapse risk.

Relapse rate and predictors among people with severe mental illnesses at Debre Markos Comprehensive specialized hospital, Northwest Ethiopia: a prospective follow up study

Severe mental illness is usually marked by periods of remission, when symptoms are absent or well controlled, and of exacerbation, when symptoms return or worsen. Relapse of these severe illnesses costs a lot for patients and their families and imposes a financial burden on hospital and community services. Costs for relapse cases were four times higher than that of non-relapse cases. There is a dearth of evidence in on relapse rate on these vulnerable population in Sub-Saharan Africa, therefore this study assessed relapse rate and predictors among people with severe mental illnesses at Debre Markos Comprehensive specialized hospital, Northwest Ethiopia. Prospective follow up study design was employed among 315 people with severe mental illnesses who were selected by systematic random sampling technique. Epi.data version 4.2 was used for data entry and exported to STATA 14 for analysis. The Kaplan-Meier curve was used to estimate the median duration of occurrence and the Log rank test was used to compare survival curves between different categories of explanatory variables. A survival analysis was used to estimate the cumulative rate of relapse, Cox proportional hazards models was used to examine independent factors associated with time to develop relapse. To estimate the association between predictors and relapse, hazard ratio with 95% confidence intervals was used. Variables score p value < 0.25 with in the Bivariable analysis was entered in to the multivariable analysis model. The statistical significance was accepted at p-value < 0.05. Around 119 (37.78%) had develop relapse, and the remaining 196 (62.22%) were censored. The overall incidence rate of relapse was 3.66 per 100 person-month (95% CI:3.06-4.38) with a total of 3250 patient-month observations. Variables such as: age (18-36 years) [(AHR) = 3.42:95% (CI) :1.67,6.97)], marital status (single and widowed) 1.87 [AHR: 1.87; 95% CI: (1.06 ,3.27)] and 2.14 [AHR: 2.14; 95% CI: (1.03 ,4.44)], duration of delay in getting treatment ( > = 1 year) [(AHR = 2.55:CI:1.20, 5.38)], types of diagnosis (Major Depressive Disorder) (AHR = 2.38, CI:1.37 ,4.14), medication adherence (low adherence) (AHR = 5.252.45, 11.21) were statistically significant (P value < 0.05). Nearly two-fifth of people diagnosis with severe mental illnesses had develop relapse and the median survival time to develop relapse was nine months. It is advised that early detection of severe mental illness and early initiation of treatments are very crucial to prevent relapse. Psycho education, counseling that alleviates poor treatment adherence are highly recommended.

Clinical characteristics associated with functioning trajectories following admission to a psychiatric institution: A prospective cohort study of individuals with psychosis

Psychotic disorders can be severely enabling, and functional recovery is often difficult to achieve. Admission to a psychiatric unit represents a key opportunity to implement strategies that will improve functional outcomes. In the current literature, there is a lack of consensus on which factors influence functional recovery. Therefore, the present longitudinal cohort study aimed to identify factors associated with functional trajectories following hospital admission for acute psychosis. A sample of 453 individuals with acute psychosis was extracted from the Signature Biobank database. Participants were followed for up to a year following admission. Various clinical indicators were documented over time. Functional trajectories were calculated based on the World Health Organization Disability Assessment Schedule 2.0. Three groups were identified: "improving", "stable", and "worsening" function. Individuals with a more severe symptomatic presentation at baseline were found to have better functional improve more over time. Over time, individuals in the "improving" and "stable" groups had significant improvements in their psychiatric symptoms. Finally, individuals following a "worsening" functional trajectory initially improved in terms of psychotic symptoms, but it did not persist over time. These results highlight the importance of studying function as a key component of recovery rather than solely focusing on relapse prevention.

Speech markers to predict and prevent recurrent episodes of psychosis: A narrative overview and emerging opportunities

Preventing relapse in schizophrenia improves long-term health outcomes. Repeated episodes of psychotic symptoms shape the trajectory of this illness and can be a detriment to functional recovery. Despite early intervention programs, high relapse rates persist, calling for alternative approaches in relapse prevention. Predicting imminent relapse at an individual level is critical for effective intervention. While clinical profiles are often used to foresee relapse, they lack the specificity and sensitivity needed for timely prediction. Here, we review the use of speech through Natural Language Processing (NLP) to predict a recurrent psychotic episode. Recent advancements in NLP of speech have shown the ability to detect linguistic markers related to thought disorder and other language disruptions within 2-4 weeks preceding a relapse. This approach has shown to be able to capture individual speech patterns, showing promise in its use as a prediction tool. We outline current developments in remote monitoring for psychotic relapses, discuss the challenges and limitations and present the speech-NLP based approach as an alternative to detect relapses with sufficient accuracy, construct validity and lead time to generate clinical actions towards prevention.

Pre-emptive ice pack cryotherapy for reducing pain caused by long-acting deltoid injectable antipsychotic treatment: A single-center open-label study

PURPOSE

This empirical study aims to investigate the efficacy of pre-emptive cryotherapy in reducing pain that is caused by the deltoid intramuscular (IM) injection of long-acting injectable (LAI) antipsychotics in clinical settings.

PATIENTS AND METHODS

This study included 29 outpatients receiving LAI antipsychotic treatment. The evaluations of pain during (1) the usual procedure (control), (2) pre-emptive use of ice pack cryotherapy (pre-cooling), and (3) pre-emptive use of a room-temperature ice pack (pre-touching) were conducted using a numerical rating scale (NRS) for comparison. All patients were administered with LAI antipsychotics via deltoid IM. Furthermore, the results of the Positive and Negative Symptom Scale (PANSS), clinical global impressions (CGI) scale, and Global Assessment of Functioning (GAF) scale that were administered during the control procedure were evaluated.

RESULTS

The median NRS pain scores during the IM injection of LAI antipsychotics were 4.0 (3.0-5.0), 2.0 (1.0-3.0), and 3.0 (2.5-6.0) for the control, pre-cooling, and pre-touching conditions, indicating a significant difference (p = 6.0 × 10-6). The NRS pain scores for the pre-cooling condition were significantly lower than those for the control and pre-touching conditions (p = 2.5 × 10-5 and 6.7 × 10-5, respectively). No significant correlation was observed between the NRS pain scores for the control condition and the PANSS, CGI scale, or GAF scale scores. Furthermore, no adverse events were recorded during the study period.

CONCLUSION

Pain during the deltoid IM injection of LAI antipsychotics was found to be reduced by pre-emptive skin cooling. To date, this is the first study to confirm the effectiveness of pre-emptive cryotherapy for relieving such pain in clinical situations.

Real-world predictors of relapse in patients with schizophrenia and schizoaffective disorder in a large health system

Schizophrenia is often characterized by recurring relapses, which are associated with a substantial clinical and economic burden. Early identification of individuals at the highest risk for relapse in real-world treatment settings could help improve outcomes and reduce healthcare costs. Prior work has identified a few consistent predictors of relapse in schizophrenia, however, studies to date have been limited to insurance claims data or small patient populations. Thus, this study used a large sample of health systems electronic health record (EHR) data to analyze relationships between patient-level factors and relapse and model a set of factors that can be used to identify the increased prevalence of relapse, a severe and preventable reality of schizophrenia. This retrospective, observational cohort study utilized EHR data extracted from the largest Midwestern U.S. non-profit healthcare system to identify predictors of relapse. The study included patients with a diagnosis of schizophrenia (ICD-10 F20) or schizoaffective disorder (ICD-10 F25) who were treated within the system between October 15, 2016, and December 31, 2021, and received care for at least 12 months. A relapse episode was defined as an emergency room or inpatient encounter with a pre-determined behavioral health-related ICD code. Patients' baseline characteristics, comorbidities and healthcare utilization were described. Modified log-Poisson regression (i.e. log Poisson regression with a robust variance estimation) analyses were utilized to estimate the prevalence of relapse across patient characteristics, comorbidities and healthcare utilization and to ultimately identify an adjusted model predicting relapse. Among the 8119 unique patients included in the study, 2478 (30.52%) experienced relapse and 5641 (69.48%) experienced no relapse. Patients were primarily male (54.72%), White Non-Hispanic or Latino (54.23%), with Medicare insurance (51.40%), and had baseline diagnoses of substance use (19.24%), overweight/obesity/weight gain (13.06%), extrapyramidal symptoms (48.00%), lipid metabolism disorder (30.66%), hypertension (26.85%), and diabetes (19.08%). Many differences in patient characteristics, baseline comorbidities, and utilization were revealed between patients who relapsed and patients who did not relapse. Through model building, the final adjusted model with all significant predictors of relapse included the following variables: insurance, age, race/ethnicity, substance use diagnosis, extrapyramidal symptoms, number of emergency room encounters, behavioral health inpatient encounters, prior relapses episodes, and long-acting injectable prescriptions written. Prevention of relapse is a priority in schizophrenia care. Challenges related to historical health record data have limited the knowledge of real-world predictors of relapse. This study offers a set of variables that could conceivably be used to construct algorithms or models to proactively monitor demographic, comorbidity, medication, and healthcare utilization parameters which place patients at risk for relapse and to modify approaches to care to avoid future relapse.

Systematic review of early warning signs of relapse and behavioural antecedents of symptom worsening in people living with schizophrenia spectrum disorders

BACKGROUND

Identification of the early warning signs (EWS) of relapse is key to relapse prevention in schizophrenia spectrum disorders, however, limitations to their precision have been reported. Substantial methodological innovations have recently been applied to the prediction of psychotic relapse and to individual psychotic symptoms. However, there has been no systematic review that has integrated findings across these two related outcomes and no systematic review of EWS of relapse for a decade.

METHOD

We conducted a systematic review of EWS of psychotic relapse and the behavioural antecedents of worsening psychotic symptoms. Traditional EWS and ecological momentary assessment/intervention studies were included. We completed meta-analyses of the pooled sensitivity and specificity of EWS in predicting relapse, and for the prediction of relapse from individual symptoms.

RESULTS

Seventy two studies were identified including 6903 participants. Sleep, mood, and suspiciousness, emerged as predictors of worsening symptoms. Pooled sensitivity and specificity of EWS in predicting psychotic relapse was 71% and 64% (AUC value = 0.72). There was a large pooled-effect size for the model predicting relapse from individual symptom which did not reach statistical significance (d = 0.81, 95%CIs = -0.01, 1.63).

CONCLUSIONS

Important methodological advancements in the prediction of psychotic relapse in schizophrenia spectrum disorders are evident with improvements in the precision of prediction. Further efforts are required to translate these advances into effective clinical innovations.

Targeted Treatment of Schizophrenia Symptoms as They Manifest, or Continuous Treatment to Reduce the Risk of Psychosis Recurrence

Current pharmacological treatment of schizophrenia employs drugs that interfere with dopamine neurotransmission, aiming to suppress acute exacerbation of psychosis and maintenance treatment to reduce the risk of psychosis recurrence. According to this treatment scheme, available psychotropic drugs intended to treat negative symptoms, cognitive impairment, or anxiety are administered as add-ons to treatment with antipsychotics. However, an alternative treatment scheme proposes a targeted or intermittent treatment approach, by which antipsychotic drugs are administered upon psychosis exacerbation and discontinued upon remission or stabilization, while negative symptoms, cognitive impairment, or anxiety are treated with specific psychotropics as monotherapy. Along these lines, antipsychotics are renewed only in the event of recurrence of psychotic symptoms. This 50-year-old debate between targeted and continuous treatment schemes arises from disagreements about interpreting scientific evidence and discordant views regarding benefit/risk assessment. Among the debate's questions are: (1) what is the percentage of individuals who can maintain stability without antipsychotic maintenance treatment, and what is the percentage of those who exacerbate despite antipsychotic treatment? (2) how to interpret results of placebo-controlled 9- to 18-month-long maintenance trials in a life-long chronic disorder, and how to interpret results of the targeted trials, some of which are open label or not randomized; (3) how to weigh the decreased risk for psychotic recurrence vs the almost certainty of adverse effects on patient's quality of life. Patients' profiles, preferences, and circumstances of the care provision should be considered as the targeted vs continuous treatment options are considered.

Health care resource utilization and costs in patients receiving long-acting injectable vs oral antipsychotics: A comparative analysis from the Disease Recovery Evaluation and Modification (DREaM) study

BACKGROUND: Given relapse frequency early in the course of schizophrenia, recently diagnosed patients may benefit from longacting injectable antipsychotics, which are associated with reduced risk of relapse and hospitalization compared with oral antipsychotics (OAPs). OBJECTIVE: To compare health care resource utilization (HCRU) and costs in patients with recent-onset schizophrenia treated with continuous paliperidone palmitate (PP) or continuous OAP or who switched from OAP to PP. METHODS: In this analysis, we combined the 2 randomized phases of the prospective, open-label Disease Recovery Evaluation and Modification (DREaM) clinical study using the principal stratification method to generate 3 treatment strategies: continuous PP for 18 months (PP-PP), continuous OAP for 18 months (OAP-OAP), and initial OAP switched to PP after 9 months (OAP-PP). HCRU metrics included psychiatric hospitalizations, psychiatric and nonpsychiatric emergency department visits, and ambulatory visits. Costs were analyzed using generalized linear models with inverse-probability weighting based on time-varying probabilities of exposure. Robust SEs were estimated using individual-level clustered bootstrapping. Subgroup analyses were performed by region and prior antipsychotic use (< 6 vs ≥ 6 months). RESULTS: A total of 181 patients were included in the PP-PP (n = 61), OAP-OAP (n = 61), and OAP-PP (n = 59) groups. The majority of patients (73%) were enrolled at study sites in the United States, and 48% had received an antipsychotic for less than 6 months prior to study entry. Baseline characteristics were well balanced, and no significant differences in discontinuation rates were observed across treatment strategies. Compared with OAP-OAP, significantly lower cumulative HCRU and costs were apparent before 9 months in the PP-PP group and after 9 months in the OAP-PP group. The cumulative 18-month effects of PP-PP and OAP-PP vs OAP-OAP on the number of psychiatric hospitalizations were ‒0.28 (95% CI = ‒0.51 to ‒0.08) and ‒0.27 (95% CI = ‒0.50 to 0.04), respectively, and those on cumulative mean per-patient total health care costs (in 2020 USD) were -$2,867 (95% CI = ‒$5,133 to ‒$750) and ‒$2,789 (95% CI = ‒$5,155 to ‒$701), respectively. Subgroup analyses indicated a greater reduction in psychiatric hospitalizations and costs with PP-PP or OAP-PP relative to OAP-OAP in patients with less than 6 vs 6 or more months of prior antipsychotic therapy. CONCLUSIONS: Continuous early use of PP in adults with recentonset schizophrenia significantly reduced psychiatric hospitalizations and associated estimated costs compared with OAP; these effects were particularly notable for patients with a shorter duration of prior antipsychotic use. As this was a post hoc analysis of a study that was not powered for HCRU assessments, future studies calibrating these effects to larger real-world populations will be useful. DISCLOSURES: Dr Basu reports consulting fees through Salutis Consulting LLC related to this work. Ms Benson, Dr Turkoz, Ms Patel, Dr Baker, and Dr Brown are employees of Janssen Scientific Affairs, LLC, and stockholders of Johnson & Johnson, Inc. This research was funded by Janssen Scientific Affairs, LLC. The sponsor was involved in the study design; collection, analysis, and interpretation of data; development and review of the manuscript; and decision to submit the manuscript for publication.