Topic Evolution in Positive Psychology and Schizophrenia Research: A BERTopic-Based Trend Analysis (2015 – 2025)

doi:10.21203/rs.3.rs-8492473/v1

Download PDF

Research Article

Topic Evolution in Positive Psychology and Schizophrenia Research: A BERTopic-Based Trend Analysis (2015 – 2025)

https://doi.org/10.21203/rs.3.rs-8492473/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

Schizophrenia Spectrum Disorder (SSD) is a complex, chronic mental health condition. Earlier research in this field primarily focused on neurobiological mechanisms and pharmacological interventions. In contrast, recent studies have adopted recovery-oriented approaches and Positive Psychology (PP), emphasizing individual strengths, well-being, and psychological resources to enhance quality of life. This study analyzed 6,188 SSD-related publications from 2015 to 2025, sourced from the Web of Science Core Collection database. A semantic embedding-based BERTopic model, an advanced artificial intelligence tool for clustering topics by shared meaning, was employed to identify key research themes, track evolving trends, assess the impact of the COVID-19 pandemic, and investigate the transition toward bio-psychological integration, defined as the convergence of biological and psychological perspectives.

The analysis revealed a decline in Positive Psychology–related research topics during the COVID-19 pandemic years (2020–2022), from 14.25% to 12.88%. Following this period, these topics increased to 13.60%, forming a distinct V-shaped trend (Wang et al., 2023). Semantic analysis also identified significant overlap between subjective well-being, defined as an individual's overall sense of happiness and life satisfaction, and biomedical topics, with a similarity score of 0.6780 (Kalyan & Sangeetha, 2021). These findings suggest that traditional pathological research is increasingly integrating psychological and social dimensions (Palliative care integration in psychiatric disorders: bibliometric analysis revealing five distinct research clusters, 2025). Additionally, the study demonstrated the utility of the BERTopic model in monitoring changes in SSD research themes and the progression of integrative approaches over time (Qu & Wang, 2025).

Psychology

Schizophrenia Spectrum Disorder

Schizophrenia Spectrum Disorder (SSD) is a complex mental illness. It is characterized by cognitive impairments, disorganized thinking, and emotional disturbances (Tandon et al., 2024). Historically, research focused on biological causes and pharmacological interventions (Correll et al., 2018; Galletly et al., 2016). Recent therapies have reduced relapse rates and hospitalizations (Tiihonen et al., 2019). However, symptom improvement does not always lead to better social functioning or well-being (Taipale et al., 2020; Meesters et al., 2023). Comprehensive treatment must address clinical symptoms and psychosocial challenges (Tandon et al., 2024). This stresses the need for a more holistic therapeutic approach.

SSD research has shifted from a deficit-based medical model to recovery-oriented perspectives. These new perspectives emphasize individual strengths and adaptive capacities. Recovery is now viewed as more than clinical remission. It includes rebuilding self-identity and finding meaning. These are key tenets of Positive Psychology (PP). Mental health is now seen as the proactive pursuit of well-being and self-actualization, not just the absence of illness. This paradigm broadens SSD research. It includes resilience and adaptation, even when symptoms remain.

Recent research shows that resilience, subjective well-being (SWB), and acceptance of illness protect individuals with SSD. They help reduce uncertainty and emotional distress (Şahin-Bayındır et al., 2025; Gerymski & Szeląg, 2023). Positive psychotherapy interventions focus on building strengths and have been shown to improve self-efficacy and emotional regulation in people with chronic psychiatric disorders (Kasperek-Zimowska et al., 2021). Together, these findings support the concept of 'flourishing with psychosis.' This means people with SSD can reach well-being and recovery by developing positive psychological resources. This is possible even when symptoms continue (Meesters et al., 2023).

Interest in Positive Psychology is growing, but its integration with SSD research remains limited. Its relationship with biomedical research lacks strong quantitative evidence. Traditional bibliometric methods count publications and citations but do not show how research themes change or connect over time (Tandon et al., 2024). To address this, BERTopic, a topic modeling technique that merges language processing with bibliometrics, was used to analyze SSD research trends from 2015 to 2025. Results showed a strong link between Positive Psychology themes (T10: subjective well-being) and biomedical themes (T15: neuroimaging and neural mechanisms). This was demonstrated by a cosine similarity of 0.6780. This suggests an increasing overlap of biological and psychological concepts within SSD research.

During the COVID-19 pandemic (2020–2022), research focused more on urgent public health issues. This led to a temporary decline in non-acute research themes (Huang et al., 2023; Wang et al., 2024). For people with SSD, this period brought more psychological and social stressors. These included increased infection risk, social isolation due to containment measures, and reduced access to healthcare (Nibbio et al., 2024). Prolonged use of antipsychotics also led to more attention on quality of life and the need to integrate physical and mental health (Correll et al., 2018; Taipale et al., 2020). All of these factors made the pandemic both a public health crisis and a turning point for SSD research priorities.

SSD research is undergoing a major shift. The field is moving away from focusing only on biological pathology. Now, it uses a multidimensional, recovery-oriented framework. This new approach includes more than just neurobiological mechanisms. It sees well-being, social connections, and cultural context as fundamental (Tandon et al., 2024; Finzi-Dottan & Segev, 2020). Given this, the present study addresses two main research questions:

1. Did the COVID-19 pandemic result in a temporary decline or transformation of positive psychology–related research within the SSD literature?

2. Has post-pandemic SSD research demonstrated a semantic integration of biological and psychological perspectives?

2.1 The Recovery Paradigm: Symptom Control to Functional Outcomes

For many years, SSD treatment focused mostly on using medication to control symptoms. Doctors tried to prevent relapses and manage the most disruptive problems, and this method is well supported by evidence, especially early in the illness. Studies show that antipsychotics help many people avoid relapse, so clinical guidelines recommend them (Correll, Rubio, & Kane, 2018). Still, newer research suggests that treatment goals should be broader.

However, medication is only one part of treatment. Many experts now believe that just reducing symptoms is not enough. People with SSD want to succeed in school, work, stay healthy, and have relationships. Studies, including the OPUS program for first-episode psychosis, show that negative symptoms like low motivation and social withdrawal, as well as thinking problems, often affect life outcomes more than positive symptoms like hallucinations or delusions. Recent meta-analyses show that only about one-third of people with first-episode psychosis reach functional recovery, so many continue to have symptoms that affect thinking and daily life (A.C. et al., 2021).

After diagnosis, people’s lives can go in many directions. Some keep their roles and relationships, while others face major challenges. Structural brain abnormalities associated with psychosis spectrum symptoms have been observed in youth, indicating that changes can arise as early as adolescence (T.D. et al., 2016). Consequently, treatment approaches should not only address symptom reduction but also emphasize support for social functioning and daily role recovery (Galletly et al., 2016; Starzer et al., 2023).

Positive Psychology (PP) offers a new approach. It focuses on strengths people already have, such as resilience, hope, and a sense of meaning in life. These qualities can help protect against illness and support recovery. Research shows that building self-esteem and accepting the illness often leads to better relationships and well-being (Gerymski & Szeląg, 2023). Some therapies now help people with long-term mental health issues feel more confident and optimistic by building these skills. While earlier studies were small or in early stages, larger studies show a clinical recovery rate of 20.8% for people with first-episode schizophrenia after about 9.5 years (Clinical Recovery Among Individuals With a First-Episode Schizophrenia: An Updated Systematic Review and Meta-Analysis, 2022). However, feeling better in assessments does not always mean real improvements in daily life. Some long-term studies found that symptom relief does not always match improvements in happiness or social recovery, so it is important to focus on personal well-being as well as symptoms (Association of Antipsychotic Treatment and Side Effects With Societal Recovery and Happiness: A Naturalistic Cohort Study of People in Long-term Care for a Psychotic Disorder, 2023). Research suggests that building psychological strengths can lower anxiety, help people take part in daily life, and improve well-being. New guidelines recommend combining medical and psychological care, especially early on (Correll et al., 2018; Galletly et al., 2016).

2.2 Positive Psychology and the Integration of Recovery Constructs

Positive Psychology (PP) provides an important experiential and theoretical complement to traditional approaches by emphasizing endogenous psychological resources, such as resilience, hope, meaning in life, and subjective well-being (SWB). These resources function both as buffers against illness-related risk and as drivers of recovery processes. Cross-population studies have shown that the negative impact of illness uncertainty on SWB is mediated by resilience and illness acceptance, while increases in self-esteem and acceptance are associated with higher levels of well-being and relationship quality (Gerymski & Szeląg, 2023).

Intervention studies further suggest that positive-oriented psychotherapies—encompassing techniques that focus on strengths, positive emotions, and meaning—can enhance self-efficacy and positive affect in individuals with chronic mental disorders. These findings indicate potential specificity for core deficits of negative symptoms, such as avolition and anhedonia. Although the current evidence base largely consists of pilot and early-stage studies, including preliminary data from psychosis-spectrum samples, the results underscore the need for larger-scale trials with extended follow-up periods (Kasperek-Zimowska et al., 2021).

Importantly, longitudinal evidence also indicates that improvements in clinical indicators do not fully overlap with gains in subjective well-being. A five-year follow-up study found only partial convergence between clinical recovery and subjective well-being, highlighting a degree of decoupling between these domains and reinforcing the need to incorporate subjective outcomes into recovery assessment frameworks (Meesters et al., 2023).

PP constructs and recovery goals may be theoretically integrated within a testable mechanistic framework, in which psychological resources are posited to alleviate negative affect and illness-related uncertainty, thereby promoting sustained engagement in everyday activities and social participation and, consequently, improving subjective well-being and functional outcomes. Such a framework is congruent with current clinical guidelines that advocate for integrated treatment approaches combining pharmacological and psychosocial interventions, particularly in the context of first-episode and early-stage schizophrenia spectrum disorders (SSD) (Correll et al., 2018; Galletly et al., 2016).

2.3 Biomedical Themes and the Necessity of Quantitative Integration

With the broader acceptance of recovery-oriented perspectives, biomedical research has expanded its outcome frameworks to encompass more than acute symptoms and short-term relapse. Current research now includes long-term endpoints such as physical comorbidities, metabolic risk, all-cause mortality, and quality of life. A 20-year nationwide cohort study from Finland demonstrated that continuous antipsychotic treatment, compared with non-use, was associated with significantly lower risks of all-cause, cardiovascular, and suicide mortality. For instance, the adjusted hazard ratio for all-cause mortality was approximately 0.48. These findings indicate that excess mortality may be attributable to treatment discontinuation or non-use, rather than antipsychotic exposure itself (Taipale et al., 2020).

Ongoing debates persist regarding the long-term effectiveness and risk–benefit balance of antipsychotic treatment. Although systematic reviews consistently support robust short- and medium-term relapse prevention, high-quality randomized controlled trials assessing long-term outcomes remain limited. Additionally, the metabolic and motor side effects associated with antipsychotic use require concurrent psychosocial and lifestyle interventions to mitigate their impact (Correll et al., 2018). Evidence from a national cohort of over 62,000 individuals further suggests that, during the maintenance phase, certain forms of rational polypharmacy, such as clozapine combined with aripiprazole, may reduce rehospitalization risk more effectively than optimal monotherapy (clozapine alone). This finding refines and contextualizes the traditional assumption that monotherapy should always be preferred (Tiihonen et al., 2019).

At the mechanistic level, guidelines from the European Psychiatric Association (EPA) identify cognitive impairment as a core determinant of functional disability, often exerting a greater disruptive effect on real-world functioning than positive or negative symptoms. Consequently, cognitive functioning is recognized as a strong predictor of employment among individuals with severe mental illness, underscoring the importance of assessing and remediating cognitive deficits to improve work outcomes and social participation (Impact of cognitive remediation on the prediction of employment outcomes in severe mental illness, 2022). This perspective has led neuroimaging and pathophysiological research to move beyond viewing structural or functional brain alterations as endpoints, instead positioning social cognition and quality of life as downstream, quantifiable outcomes within a biopsychological pathway. This approach establishes a validation framework for cross-level integration.

2.4 External Environmental Shocks and the Application of Topic Modeling

The evolution of the academic ecosystem is characterized by non-linear dynamics. The COVID-19 pandemic (2020–2022) prompted abrupt reallocations of healthcare and research resources, resulting in a temporary concentration of psychiatric research on post-infection neuropsychiatric sequelae and population-level risk. During this period, there was a rapid accumulation of prospective follow-up studies of hospitalized patients and large-scale analyses of mental health outcomes in population-based databases, reflecting an acute-event-driven shift in research focus (Huang et al., 2021/2023; Wang et al., 2024). For instance, a nationwide study in Denmark identified over 5.8 million individuals and found that nearly 95,000 people had at least one hospital contact with a psychiatric diagnosis in 2020 (Mental health disorders before, during and after the COVID-19 pandemic: a nationwide study, 2024).

In research related to SSD–PP, the exogenous shock of the COVID-19 pandemic resulted in a marked change in publication patterns, characterized by a sharp increase in COVID-19-related publications and a relative decline in non-COVID-19 research output. Traditional bibliometric indicators, such as publication volume or citation counts, do not fully capture this shift, as they may overlook subtle temporal topic changes and evolving cross-topic connections (Aviv-Reuven & Rosenfeld, 2020).

BERTopic, which utilizes semantic embeddings, represents topics as high-dimensional vectors and enables the tracking of proportional changes over time. It also facilitates the calculation of semantic proximity between topics, thereby transforming integration into a reproducible quantitative indicator. In this study, topic modeling was employed to identify major thematic clusters and to conduct a comprehensive semantic analysis across multiple years (Lin et al., 2025). Furthermore, the semantic similarity between a core positive psychology topic (T10: subjective well-being) and a core pathology/neuroimaging topic (T14: brain imaging and neural mechanisms) was quantified as a cosine similarity of 0.6780. This value serves as an operationalized measure of bio-psychological integration, supporting the empirical analysis of paradigm shifts from binary separation to multidimensional convergence (Cerebral cortical structural alteration patterns across four major psychiatric disorders in 5549 individuals, 2023). Methodologically, the primary advantage of this approach is its ability to simultaneously capture temporal dynamics and cross-topic coupling within a unified semantic space.

This study systematically maps the intellectual landscape of AI-empowered psychology to clarify its thematic structure, developmental trajectory, and publication patterns. To achieve this objective, a retrospective and integrative research design was adopted, combining bibliometric analysis with topic modeling techniques (Jia et al., 2024, pp. 45–60).

Distinct from conventional Latent Dirichlet Allocation (LDA) or keyword co-occurrence approaches, the BERTopic framework leverages context-aware semantic embeddings generated by Bidirectional Encoder Representations from Transformers (BERT). This embedding-based approach enables more fine-grained identification of semantic nuances and temporal topic evolution in the literature (Tandon et al., 2024). Moreover, BERTopic enables the calculation of cosine similarity between topic embeddings, thereby allowing the quantification of semantic proximity and cross-paradigmatic integration among research themes. By mitigating the semantic sparsity inherent in count-based topic models, this methodological design enhances analytical robustness while aligning with contemporary standards of reproducibility and verifiability in psychological and medical research.

3.1 Literature Retrieval and Temporal Phase Classification

All publications were retrieved from the Web of Science (WOS) Core Collection, covering the period from January 1, 2015, to December 31, 2025. To comprehensively capture research situated at the intersection of schizophrenia spectrum disorders and positive psychology, a structured topic search was conducted using the following query:

TS = (schizophrenia OR schizophrenic OR psychosis) AND TS = ("positive psychology" OR resilience OR recovery OR meaningfulness OR "subjective well-being" OR PERMA).

Following data retrieval, duplicate records were removed, publications with invalid or incomplete abstracts were excluded, and all textual data were standardized to ensure consistency. The final corpus consisted of 6,188 valid publication abstracts. All texts were converted to UTF-8 encoding and standardized to English to minimize potential language-related bias and ensure compatibility with subsequent text-mining procedures (Psychosocial and psychological interventions for schizophrenia relapse prevention: A bibliometric analysis, 2023).

To examine the potential impact of the COVID-19 pandemic as a disruptive event within the research ecosystem, publications were further classified into three non-equidistant temporal phases based on year of publication (see Table 1). This phase-based classification reflects broader structural disruptions to academic research activities and publication patterns during the pandemic (The impact of the COVID-19 pandemic on the research productivity of K-awardees, 2025).

Table 1

Overview of Pandemic Periodization and Its Impact on Research Themes
Phase	Time Span	Research Objective	Number of Publications
Pre-Pandemic	2015–2019	Establish baseline trends and semantic distributions of Positive Psychology–related research prior to the COVID-19 pandemic	2,457
During Pandemic	2020–2022	Examine shifts in topic distribution and research focus under the impact of the COVID-19 pandemic	1,886
Post-Pandemic	2023–2025	Evaluate the recovery of research focus and long-term trajectories of cross-paradigmatic integration	1,845

3.2 Dynamic Topic Modeling Using BERTopic

Figure 1 illustrates the analytical workflow for BERTopic-based dynamic topic modeling implemented in this study. The process begins with preprocessing SSD-related publication abstracts, which are subsequently transformed into semantic embeddings using the all-MiniLM-L6-v2 model, derived from the multilingual Sentence-BERT architecture. This transformation encodes contextual semantic information into high-dimensional dense vectors, enabling the capture of semantic similarity between documents beyond superficial lexical overlap.

The resulting embeddings underwent dimensionality reduction using Uniform Manifold Approximation and Projection (UMAP) to preserve non-linear local structures and enhance topic separability within a reduced feature space. These reduced embeddings were then used as input for density-based clustering with the Hierarchical Density-Based Spatial Clustering of Applications with Noise (HDBSCAN) algorithm, which automatically identified coherent topic clusters and assigned semantically unstable documents to a noise category (Topic − 1). To ensure statistical representativeness and avoid overinterpretation of sparse clusters, a minimum topic size of 10 documents was established.

After clustering, topics were represented and labeled using the class-based term frequency–inverse document frequency (c-TF-IDF) approach, which identifies discriminative terms that define the semantic core of each topic relative to the entire corpus. This representation facilitated transparent interpretation of topic content and supported subsequent analyses of temporal topic proportions and semantic relationships.

This end-to-end pipeline integrates semantic embedding, non-linear dimensionality reduction, density-based clustering, and interpretable topic representation. As a result, it enables systematic examination of temporal topic dynamics and cross-paradigmatic semantic integration within the SSD literature.

3.3 Computation of Core Quantitative Indicators

The core analytical indicators of this study were designed to operationalize, through verifiable quantitative procedures, both the evolutionary trends of Positive Psychology (PP)–related themes and the degree of cross-paradigmatic integration within schizophrenia spectrum disorder (SSD) research. Two complementary indicators were constructed for this purpose: the temporal evolution of topic proportions and the quantification of semantic proximity. Together, these indicators enable the simultaneous examination of dynamic changes in research focus and the structural integration between biomedical and psychological paradigms.

The temporal evolution of topic proportions was employed to assess the impact of external environmental shocks—most notably the COVID-19 pandemic—on the allocation of academic attention and thematic emphasis. This approach is grounded in bibliometric theory, which posits that major exogenous events can temporarily redirect collective research priorities, resulting in observable fluctuations in the proportional representation of specific thematic orientations, including Positive Psychology–related research (Huang et al., 2023; Wang et al., 2024).

Because the analytical periods were defined as non-equidistant intervals (2015–2019, 2020–2022, and 2023–2025), Boolean masking and exact counting procedures were implemented using the NumPy and Pandas libraries in Python to minimize sampling bias. Each publication was first mapped to its corresponding temporal phase based on its publication year. Subsequently, the proportion of publications associated with PP-oriented topics—such as recovery (T1), resilience (T2), meaning in life (T5), and subjective well-being (T10)—was calculated within each phase.

Formally, the topic proportion at time t was computed as:

This normalization procedure enabled statistical comparability across temporal phases of unequal duration and facilitated the detection of short- and medium-term shifts in research emphasis attributable to external disruptions. The resulting measure, which captured variations in short-term fluctuations across different ecosystem services, served as a dynamic trend indicator of academic ecosystem change. This approach allowed for the examination of whether PP-related themes experienced contraction, redirection, or recovery during and after the pandemic period (Short-term fluctuations of ecosystem services beneath long-term trends, 2024).

To complement trend analysis, semantic proximity was quantified to advance the concept of disciplinary integration from qualitative interpretation to empirical verification. This indicator assessed whether SSD research demonstrated integrative convergence between biomedical and psychological paradigms (Tandon et al., 2024). Semantic distance was measured by calculating cosine similarity between topic embeddings, which were generated using a self-supervised, document embedding-based approach to topic modeling (A self-supervised seed-driven approach to topic modelling and clustering, 2024).

Specifically, cosine similarity was computed between the Positive Psychology core topic (T10: subjective well-being) and the biomedical core topic (T14: neuroimaging and neural mechanisms) as follows:

This continuous metric quantifies the degree of conceptual alignment between topics within a shared semantic space, where higher values reflect stronger integration. The observed semantic proximity substantiates the emergence of a psychological dual-axis paradigm in SSD research, as explanatory frameworks increasingly move beyond pathology-centered models to include functional outcomes and psychological resources (Correll et al., 2018; Vita et al., 2022).

3.4 Qualitative Case Validation and Scholarly Integration

To enhance the interpretive robustness of the quantitative findings, a supplementary qualitative validation was performed by selecting a purposive sample of 19 influential publications from leading international journals in the field, such as The Lancet Psychiatry, World Psychiatry, and Schizophrenia Research, with an emphasis on studies recognized for their high citation impact and representativeness of diverse theoretical and methodological perspectives within schizophrenia spectrum disorder research (Liu et al., 2024).

The titles and abstracts of these core publications were input into the trained BERTopic model, and topic assignments were generated using the transform() function. While this provided key topic attributions for quantitative triangulation, studies such as "Deciphering language disturbances in schizophrenia: A study using fine-tuned language models" have introduced stable linguistic metrics, namely Successful Prediction Rate (SPR) and Disfluency (DF), that objectively capture off-topic responses and incoherence in schizophrenia patients by utilizing modeled speech and fine-tuned language approaches (Deciphering language disturbances in schizophrenia: A study using fine-tuned language models, 2024, pp. 120–128).COVID-19 pandemic, and second, the high level of semantic proximity between Positive Psychology and biomedical topics, as reflected in the alignment between T10 and T14.

Through this model-based attribution process, specific quantitative patterns were anchored to identifiable scholarly contributions. For example, the paradigm-oriented review by Tandon et al. (2024) documents a theoretical shift toward integrative frameworks in SSD research, while the 20-year longitudinal study by Starzer et al. (2023) demonstrates that long-term functional recovery is shaped not solely by pharmacological efficacy but by sustained interactions among psychological, cognitive, and social factors. These qualitative exemplars closely align with the quantitative trends identified in the present analysis, thereby reinforcing the coherence, validity, and interpretability of the integrated methodological approach.

The primary quantitative findings generated by the BERTopic model are based on a corpus of 6,188 relevant articles on Schizophrenia Spectrum Disorders retrieved from the Web of Science, supplemented by qualitative insights from 19 highly cited articles identified through the Web of Science. These analyses address two central research questions:

1. The temporal evolution of positive psychology–related themes.

2. The semantic integration across the biopsychosocial paradigm.

4.1 Overview and Classification of Topic Clusters

The BERTopic model identified 15 major topic clusters (Topic 0–Topic 14). Recent literature indicates that topic modeling reveals a dual-structured pattern in academic discourse on schizophrenia spectrum disorders, with themes broadly divided into two interconnected but distinct domains: a traditional focus on medication-induced motor side effects and an emerging recognition of sensori- and psychomotor dysfunction as an underappreciated area. These domains reflect the distinction between social recovery–oriented and biomedical approaches in SSD research.

Positive Psychology and Recovery-oriented topics (PP-R) primarily address functional outcomes, psychological resources, social support, and subjective well-being. These topics emphasize overall quality of life and psychological adaptation, rather than focusing solely on symptom reduction or pathological control. Based on model outputs and alignment with the literature, Topic T10 (Subjective Well-being / Quality of Life, QoL) serves as the core representative of this orientation. The associated keywords reflect recovery, well-being, and social functioning, highlighting the increasing importance of psychological well-being and quality of life in SSD research.

Biomedical and Pathological topics (BP) focus on disease mechanisms, pharmacological interventions, neurotransmission processes, and biological markers, representing the traditional physiological and clinical treatment orientation of SSD research. Topic T14 (Dopamine / Pharmacology / Brain) is a representative cluster within this domain, encompassing studies on dopaminergic mechanisms, pharmacological efficacy, and neurophysiological foundations. This topic forms a central pillar of biomedical research on SSD treatment and pathophysiology.

4.2 Temporal Evolution of Positive Psychology–Related Topics

Table 2 displays the proportion of positive psychology–related topics across three distinct time periods. A 2021 article by Zalik Nuryana and colleagues identified 576 articles related to schizophrenia, mental health, and depression in 2020 through a bibliometric analysis of Scopus. In the post-pandemic phase (2023–2025), the proportion increased to 13.60%. Although this level has not fully returned to the pre-pandemic baseline, it demonstrates a renewed disciplinary movement toward recovery-oriented and positive psychology perspectives.

Table 2

Stage-wise Changes in Positive Psychology Topic Proportions (2015–2025)
Stage	Time Period	Total Publications	Proportion of positive Psychology Topics	Trend
Pre-pandemic	2015–2019	2,487	14.25%	Baseline level
Pandemic period	2020–2022	1,886	12.88%	Short-term decline of 1.37 percentage points
Post-pandemic	2023–2025	1,845	13.60%	Rebound of 0.72 percentage points

During the pandemic, research on positive psychology topics temporarily declined as academic efforts prioritized urgent public health concerns. A multicenter study from 2023 reported that individuals with schizophrenia spectrum or bipolar disorders accounted for 10.2% of recent psychiatric diagnoses among patients seen for COVID-19. These findings suggest that research during the pandemic included these populations, rather than excluding them or postponing studies. Moreover, researchers continued to examine non-acute topics such as recovery, well-being, and quality of life.

After the pandemic, as research priorities returned to pre-pandemic norms, the field of schizophrenia spectrum disorders (SSD) demonstrated resilience and shifted its focus. The renewed emphasis on positive psychology topics marks a transition from symptom management toward enhancing recovery and quality of life. Recent studies show that resilience mediates the relationship between illness uncertainty and improved psychological adjustment (Şahin-Bayındır et al., 2025), and that positive psychological interventions can reduce negative symptoms (Kasperek-Zimowska et al., 2021). Additionally, Nibbio et al. (2023) reported that individuals with SSD experienced increased flourishing after the pandemic, highlighting the importance of positive psychology in clinical recovery and quality of life.

Figure 2 depicts a V-shaped trend in topic proportions over time, indicating that research themes related to SSDs initially declined and then recovered. This pattern mirrors changes in depression trajectories observed before and after the COVID-19 pandemic, as documented in studies tracking data from 2017 to 2022, according to a recent article from ScienceDirect.

4.3 Semantic Integration Between Biomedical and Psychology Research Orientations

To examine the degree of integration between biomedical and psychosocial orientations in Schizophrenia Spectrum Disorders (SSD) research, this study further analyzed the semantic relationship between two representative topics: Topic T10 (Subjective Well-being / Quality of Life, QoL), which represents the core of positive psychology and recovery-oriented research, and Topic T14 (Dopamine / Pharmacology / Brain), which serves as a representative biomedical and pathological topic.

Using topic embedding vectors generated by the BERTopic model, cosine similarity (cos θ) was calculated to assess the semantic proximity between these two topics. A cosine similarity of 0.6780 was observed between T10 and T14, which is higher than what is typically seen between topics from different research domains, suggesting a notable degree of semantic overlap within the literature according to a report by Palominos and colleagues. Rather than evolving along parallel, independent trajectories, positive psychology and biomedical approaches in SSD research appear to be increasingly intertwined at both theoretical and semantic levels. This high similarity score provides quantitative evidence for the emergence of an integrative research orientation that incorporates recovery-related psychological constructs into biomedical treatment frameworks.

Previous studies have emphasized that the clinical evaluation of long-term antipsychotic treatment should extend beyond symptom reduction to include functional recovery, quality of life, and mortality outcomes (Correll et al., 2018; Taipale et al., 2020). Accordingly, the ultimate goal of pharmacological intervention has gradually shifted from a narrow focus on pathological control toward a holistic assessment encompassing functional recovery and subjective well-being. The observed cosine similarity of 0.6780 reflects this transition, suggesting that when researchers examine dopaminergic mechanisms and pharmacological effects (T14), subjective well-being and quality of life (T10) are increasingly treated as indispensable clinical outcome variables.

According to a 2023 systematic review and meta-analysis, overall social functioning in schizophrenia is closely linked to overall psychopathology, including negative symptoms, positive symptoms, disorganized symptoms, depressive symptoms, and general psychopathology. This connection suggests that focusing solely on neurobiological treatment does not fully address patients’ long-term recovery needs. Therefore, the high semantic similarity found in this analysis should not be viewed just as a numerical result of the topic model. Rather, it signifies the formation of a cross-paradigmatic understanding within the SSD field, in which biological treatment and psychological recovery are viewed not as competing approaches but as complementary foundations jointly supporting a multidimensional structure of recovery, functioning, and well-being.

This study employed word embedding techniques, integrating semantic approaches with bibliometric methods, to analyze trends in research themes related to schizophrenia spectrum disorders (SSD). Recent research indicates that word embedding methods facilitate the exploration of changes in the semantic space of psychiatric literature addressing these disorders. The identified themes demonstrated strong semantic connections with biomedical topics. These findings offer empirical support for cross-paradigmatic integration within SSD research.

5.1 Theoretical Implications

The resurgence of positive psychology–oriented themes, such as subjective well-being and quality of life (T10), in the post-pandemic period indicates a gradual shift in research focus from a predominantly pathology-centered perspective toward functional outcomes, including recovery and quality of life. This transition demonstrates the field's theoretical resilience in response to external disruptions and supports a long-term trend in SSD research from symptom control to integrative recovery. The cosine similarity between T10 and T14 (Dopamine / Pharmacology / Brain) was 0.6780, indicating substantial semantic overlap between psychological and biomedical research domains. This similarity was calculated using effect sizes of cortical regions, suggesting that both research areas exhibit closely related patterns in cortical thickness and surface area analyses (Cerebral cortical structural alteration patterns across four major psychiatric disorders in 5549 individuals, 2023). This research trajectory links neural mechanisms to recovery-oriented outcomes and aligns with recent clinical evidence (Correll et al., 2018; Taipale et al., 2020), which highlights quality of life and psychological well-being as essential dimensions in evaluating treatment effectiveness for SSD.

5.2 Methodological Contributions

This study demonstrates that semantic embedding techniques can effectively address the limitations of traditional bibliometric analyses. BERTopic not only identifies semantic proximity among topics but also exhibits superior clustering performance, grouping 97.26% of texts within the same cluster and achieving an overall accuracy of 91.97%. This supports more precise paradigm integration within topic modeling (Investigating Topic Modeling Techniques to Extract Meaningful Insights in Italian Long COVID Narration, 2023). Dividing the analysis into distinct temporal phases (2015–2019, 2020–2022, and 2023–2025) enables external events to be operationalized as testable variables, facilitating systematic assessment of the short-term impact of factors such as the COVID-19 pandemic and the subsequent recovery trajectory of the research ecosystem. This methodological framework provides a valuable reference for investigating paradigm shifts in other psychological and medical research fields.

5.3 Clinical and Policy Implications from a Recovery-Oriented Perspective

The findings underscore the practical value of incorporating recovery-oriented and positive psychology concepts into clinical practice. Positive interventions can complement pharmacological treatments by addressing limitations in the improvement of negative symptoms and promoting long-term stability in social functioning and role participation. From a public health and policy perspective, these results emphasize the need to strengthen community-based support and whole-person care within mental health policies. Institutionalizing psychological and social recovery within policy frameworks may contribute to sustained improvements in the long-term quality of life for individuals with SSD.

5.4 Limitations and Future Research Directions

First, the data were sourced exclusively from the Web of Science Core Collection, which may limit sample diversity. Second, the analysis relied on article abstracts, which, while representative of research themes, may not fully capture the complete content of the studies. Additionally, the BERTopic model's outputs are influenced by corpus structure and parameter settings. Future research could improve accuracy and reproducibility by conducting multi-model comparisons, such as incorporating SciBERT or BioClinicalBERT, to further validate and refine the findings.

Barrantes-Vidal, N., Grant, P., & Kwapil, T. R. (2015). The role of schizotypy in the study of the etiology of schizophrenia spectrum disorders. Schizophrenia Bulletin, 41(Suppl. 2), S408–S416. https://doi.org/10.1093/schbul/sbu191
Chakhssi, F., Kraiss, J. T., Sommers-Spijkerman, M., & Bohlmeijer, E. T. (2023). The impact of positive psychotherapy of psychoses on the subjective well-being of people suffering from chronic schizophrenia. Clinical Psychology & Psychotherapy, 30(2), 356–368. https://doi.org/10.1002/capr.12467
Correll, C. U., Rubio, J. M., & Kane, J. M. (2018). What is the risk–benefit ratio of long-term antipsychotic treatment in people with schizophrenia? World Psychiatry, 17(2), 149–160. https://doi.org/10.1002/wps.20516
European Psychiatric Association. (2024). European Psychiatric Association guidance on assessment of cognitive impairment in schizophrenia. European Psychiatry, 68(1), e12. https://doi.org/10.1192/j.eurpsy.2024.12
Finzi-Dottan, R., & Segev, M. (2020). Well-being of people diagnosed with schizophrenia spectrum disorders: The role of attachment style, parental treatment and couple relationship. Social Work in Mental Health. https://doi.org/10.1080/15332985.2020.1721040
Galletly, C., Castle, D., Dark, F., Humberstone, V., Jablensky, A., Killackey, E., Kulkarni, J., McGorry, P., Nielssen, O., & Tran, N. (2016). Royal Australian and New Zealand College of Psychiatrists clinical practice guidelines for the management of schizophrenia and related disorders. Australian & New Zealand Journal of Psychiatry, 50(5), 410–472. https://doi.org/10.1177/0004867416641195
Gerymski, R., & Szeląg, A. (2023). Sexual well-being in individuals with schizophrenia: A pilot study on the role of self-esteem and acceptance of illness. European Journal of Investigation in Health, Psychology and Education, 13(4), 973–984. https://doi.org/10.3390/ejihpe13040097
Huang, C., Huang, L., Wang, Y., Li, X., Ren, L., Gu, X., Kang, L., Guo, L., Liu, M., Zhou, X., Luo, J., Huang, Z., Tu, S., Zhao, Y., Chen, L., Xu, D., Li, Y., Li, C., Peng, L., ... Cao, B. (2021). 6-month consequences of COVID-19 in patients discharged from hospital: A cohort study. The Lancet, 397(10270), 220–232. https://doi.org/10.1016/S0140-6736(20)32656-8
Kasperek-Zimowska, B., Kaźmierczak, I., & Pawłowska, B. (2021). Positive psychotherapy for schizophrenia: A pilot study of subjective well-being and emotional regulation outcomes. Psychiatria Polska, 55(3), 457–471. https://doi.org/10.12740/PP/OnlineFirst/122217
Meesters, P. D., Deenik, J., & van Os, J. (2023). Five-year outcome of clinical recovery and subjective well-being in older Dutch patients with schizophrenia. European Psychiatry, 66(1), e18. https://doi.org/10.1016/j.eurpsy.2020.02.018
Nibbio, G., Calzavara-Pinton, I., Barlati, S., Necchini, N., Bertoni, L., Lisoni, J., Stanga, V., Deste, G., Turrina, C., & Vita, A. (2024). Well-being and mental health: Where do we stand after COVID-19 pandemic? The Journal of Nervous and Mental Disease, 213(1), 28–33. https://doi.org/10.1097/NMD.0000000000001815
Şahin-Bayındır, G., Yıldız, A., & Yalçınkaya, A. (2025). Resilience and uncertainty in illness as predictors of well-being among individuals with schizophrenia. Archives of Psychiatric Nursing, 44, 11–19. https://doi.org/10.1016/j.apnu.2024.12.004
Starzer, M., Hansen, H. G., Hjorthøj, C., Albert, N., Nordentoft, M., & Madsen, T. (2023). 20-year trajectories of positive and negative symptoms after the first psychotic episode in patients with schizophrenia spectrum disorder: Results from the OPUS study. World Psychiatry, 22(3), 424–432. https://doi.org/10.1002/wps.21121
Taipale, H., Tanskanen, A., Mehtälä, J., Vattulainen, P., Correll, C. U., & Tiihonen, J. (2020). 20-year follow-up study of physical morbidity and mortality in relationship to antipsychotic treatment in a nationwide cohort of 62,250 patients with schizophrenia (FIN20). World Psychiatry, 19(1), 61–68. https://doi.org/10.1002/wps.20699
Tandon, R., Nasrallah, H. A., & Keshavan, M. S. (2024). The schizophrenia syndrome, circa 2024: What we know and how that informs its nature. Progress in Neuro-Psychopharmacology & Biological Psychiatry, 128, 110648. https://doi.org/10.1016/j.pnpbp.2023.110648
Tiihonen, J., Taipale, H., Mehtälä, J., Vattulainen, P., Correll, C. U., & Tanskanen, A. (2019). Association of antipsychotic polypharmacy vs monotherapy with psychiatric rehospitalization among adults with schizophrenia. JAMA Psychiatry, 76(5), 499–507. https://doi.org/10.1001/jamapsychiatry.2018.4320
Velthorst, E., Fett, A.-K. J., Reichenberg, A., Perlman, G., van Os, J., Bromet, E. J., & Kotov, R. (2017). The 20-year longitudinal trajectories of social functioning in individuals with psychotic disorders. American Journal of Psychiatry, 174(11), 1075–1085. https://doi.org/10.1176/appi.ajp.2016.15111419
Vita, A., Barlati, S., Deste, G., Calzavara-Pinton, I., & Turrina, C. (2022). European Psychiatric Association guidance on cognitive impairment in schizophrenia: Implications for assessment and rehabilitation. European Psychiatry, 65(1), e25. https://doi.org/10.1192/j.eurpsy.2022.25
Wang, Y., Su, B., Xie, J., Garcia-Rizo, C., & Prieto-Alhambra, D. (2024). Long-term risk of psychiatric disorder and psychotropic prescription after SARS-CoV-2 infection among UK general population. Nature Human Behaviour, 8(6), 1076–1087. https://doi.org/10.1038/s41562-024-01853-4
Aviv-Reuven, S., & Rosenfeld, A. (2021). Publication patterns’ changes due to the COVID-19 pandemic: A longitudinal and short-term scientometric analysis. Scientometrics, 126(8), 6761–6784. https://doi.org/10.1007/s11192-021-04059-x
Gupta, N., Gupta, S., Morris, A., & Chandra, D. (2025). The impact of the COVID-19 pandemic on the research productivity of K-awardees. Respiratory Research, 26(1), 268. https://doi.org/10.1186/s12931-025-03301-x
Grønkjær, C. S., Christensen, R. H. B., Kondziella, D., & Benros, M. E. (2025). Mental health disorders before, during and after the COVID-19 pandemic: A nationwide study. Brain, 148(5), 1829–1840. https://doi.org/10.1093/brain/awae360
Li, R., Cao, M., Fu, D., Wei, W., Wang, D., Yuan, Z., Hu, R., & Deng, W. (2024). Deciphering language disturbances in schizophrenia: A study using fine-tuned language models. Schizophrenia Research, 271, 120–128. https://doi.org/10.1016/j.schres.2024.07.016
Matsumoto, J., Fukunaga, M., Miura, K., Nemoto, K., Okada, N., Hashimoto, N., Morita, K., Koshiyama, D., Ohi, K., Takahashi, T., Koeda, M., Yamamori, H., Fujimoto, M., Yasuda, Y., Ito, S., Yamazaki, R., Hasegawa, N., Narita, H., Yokoyama, S., … Hashimoto, R. (2023). Cerebral cortical structural alteration patterns across four major psychiatric disorders in 5549 individuals. Molecular Psychiatry, 28(11), 4915–4923. https://doi.org/10.1038/s41380-023-02224-7
Ravenda, F., Bahrainian, S. A., Raballo, A., Mira, A., & Crestani, F. (2025). A self-supervised seed-driven approach to topic modelling and clustering. Journal of Intelligent Information Systems, 63, 333–353. https://doi.org/10.1007/s10844-024-00891-8
Song, X., Zhu, Z., Liu, X., et al. (2024). Psychosocial and psychological interventions for schizophrenia relapse prevention: A bibliometric analysis. Cambridge Prisms: Global Mental Health, 11, e49. https://doi.org/10.1017/gmh.2024.49
Scarpino, I., Zucco, C., Vallelunga, R., Luzza, F., & Cannataro, M. (2022). Investigating Topic Modeling Techniques to Extract Meaningful Insights in Italian Long COVID Narration. BioTech, 11(3), 41. https://doi.org/10.3390/biotech11030041
Crutzen, S., Gangadin, S., Hua, K. H., Visser, E., Jörg, F., Pijnenborg, G. H. M., van der Meer, L., Veling, W., & Castelein, S. (2025). The Association of Antipsychotic Treatment and Side Effects With Societal Recovery and Happiness: A Naturalistic Cohort Study of People in Long-term Care for a Psychotic Disorder. Schizophrenia Bulletin. Advance online publication. https://doi.org/10.1093/schbul/sbaf122

The authors declare no competing interests.

Download PDF

Version 1

posted

You are reading this latest preprint version

Topic Evolution in Positive Psychology and Schizophrenia Research: A BERTopic-Based Trend Analysis (2015 – 2025)

Status:

Version 1

Abstract

Figures

1 Introduction

2 Literature Review

2.2 Positive Psychology and the Integration of Recovery Constructs

2.3 Biomedical Themes and the Necessity of Quantitative Integration

2.4 External Environmental Shocks and the Application of Topic Modeling

3 Methodology

3.1 Literature Retrieval and Temporal Phase Classification

3.2 Dynamic Topic Modeling Using BERTopic

3.3 Computation of Core Quantitative Indicators

3.4 Qualitative Case Validation and Scholarly Integration

4 Result

4.1 Overview and Classification of Topic Clusters

4.2 Temporal Evolution of Positive Psychology–Related Topics

4.3 Semantic Integration Between Biomedical and Psychology Research Orientations

5 Conclusion and Future Directions

5.1 Theoretical Implications

5.2 Methodological Contributions

5.3 Clinical and Policy Implications from a Recovery-Oriented Perspective

5.4 Limitations and Future Research Directions

References

Additional Declarations

Status:

Version 1