Published papers

Mental health is socially an important issue in society. In recent years, the mental health issue has been closely related to activities on the Internet. We have developed a system for classifying Twitter users whether they have mental health issues. We assumed that accounts that match specific patterns are likely to have mental health issues, and collected positive and negative examples based on this assumption. In order to investigate the possibility to classify user utterances without mental health specific keywords, we discarded tweets which include such keywords, in contrast to keyword dependent previous studies. We trained and compared classification performances of BERT, LSTM, and SVM. Regarding BERT, we originally pre-trained the model using the large-scale tweet data. BERT achieved high classification performance with Accuracy 0.83, Recall 0.8, Precision 0.88, and F1-score 0.84 in the best performance case.

© 2020 International Parkinson and Movement Disorder Society Background: Electroconvulsive therapy (ECT) is a well-established treatment for psychiatric disorders, including depression and psychosis. ECT has been reported to be effective in treating such psychiatric symptoms in patients with Parkinson's disease (PD) and has been also reported to be effective in treating motor symptoms. The aim of the study is to summarize previous clinical studies investigating the efficacy of ECT for symptoms in patients with PD. Methods: A systematic review and meta-analysis of any study designs assessing motor and/or non-motor symptoms in patients with PD before and after ECT. Co-primary outcomes were set as motor manifestations assessed using the Unified Parkinson's Disease Rating Scale or other rating scales, and non-motor symptoms included depression and psychosis. Secondary outcomes were wearing-off phenomenon and cognitive function. The impact of ECT on those symptoms was examined by comparing the severity of the symptoms before and after ECT using a random effect model and was expressed in standardized mean difference. Results: Of 1219 identified citations, 14 studies (n = 129; 1 randomized controlled study, 9 prospective observational studies, and 4 retrospective studies) were analyzed. The findings were as follows: ECT significantly improved motor manifestations in patients with PD, and the improvement was significant in the subpopulation without psychiatric symptoms; ECT significantly improved depression and psychosis; and ECT significantly relieved wearing-off phenomenon and did not worsen cognitive functioning. Conclusion: The current meta-analysis suggests the potential benefit of ECT on motor and non-motor symptoms in presumably complicated and difficult-to-treat subgroups. © 2020 International Parkinson and Movement Disorder Society.

Background: The use of telemedicine is expanding worldwide in response to the COVID-19 pandemic, and deregulation is occurring even in countries with strict regulations, such as Japan. Methods: We conducted a narrative review of the activities and initiatives undertaken by the government and professional organizations to promote the use of telemedicine in Japan from January 2020, when the COVID-19 pandemic began, to September 2021. Results: In Japan, where before the pandemic the spread of telemedicine had been limited, various recommendations and deregulations have now been made, resulting in the expanded use of telemedicine. In particular, the government's overall policy direction has become a catalyst for deregulation. Conclusions: With the COVID-19 pandemic, telemedicine is now moving toward widespread use. However, further activities and initiatives, including policy changes such as reimbursement and the establishment of evidence to show the effectiveness of telemedicine in Japan, are needed to properly disseminate telemedicine.

Objective Electroconvulsive therapy (ECT) is provided in real-world clinical settings for patients lacking capacity for consent. The aim of this study was to investigate the clinical characteristics and clinical effectiveness of ECT in this population. Methods A retrospective chart review was conducted to collect data from patients who received ECT to treat their depressive episodes between April 2012 and March 2019. Differences in clinical characteristics and short-/long-term clinical outcomes between patients who received ECT with their relatives' consent and patients who received ECT by their own consent were examined. The short-/long-term clinical outcomes were determined by clinical global impression scores and readmission rate, respectively. Results Of 168 patients with depressive episodes, 34 (20.2%) received ECT with their relatives' consent. Those patients were older, had lower body mass index, and had shorter episode duration. They also exhibited more frequent psychotic, melancholic, and catatonic features. The main indication for ECT in this population was the need for rapid recovery. Patients lacking capacity for consent showed similar remission (61.8%) and response (82.4%) rates to those with capacity for consent. Readmission rate was not significantly different between groups. Conclusions There were no significant differences in short-/long-term ECT effectiveness between patients with/without capacity for consent. Electroconvulsive therapy is the only established and effective treatment in clinical settings for the most severe cases, wherein patients are incapable of giving consent but need rapid recovery. A general rejection of this practice due to concerns surrounding consent may be unethical under the ethical principles of medical care.

Background: Evidence of comparative benefits of long-acting injectable antipsychotics (LAIs) versus oral antipsychotics for schizophrenia has been inconsistent across study designs. The aim of this study was to evaluate the comparative benefits of LAIs versus oral antipsychotics in three study designs to inform clinical decision making. Methods: We did a comprehensive systematic review and meta-analysis comparing LAIs versus oral antipsychotics for schizophrenia covering three study designs: randomised controlled trials (RCTs), cohort studies, and pre–post studies. Our literature search was without language restrictions, in MEDLINE and PubMed, the Cochrane Library, Scopus, and Embase, for studies published from database inception up to a last search on March 13, 2020. We also searched for unpublished studies and ClinicalTrials.gov. We included studies lasting at least 6 months that targeted adults with schizophrenia and related disorders (>80% of participants). Studies on penfluridol (neither an LAI or daily oral antipsychotic), case reports, and case series with fewer than 20 patients were excluded. Two investigators independently extracted study-level data and resolved disagreement by consensus, or via a third investigator. Study authors were contacted to obtain additional information as needed. For our primary outcome we meta-analysed the risk ratio (RR) for hospitalisation or relapse with LAIs versus oral antipsychotics by a random-effects model, with hospitalisation used preferentially over relapse. As secondary analyses, we reversed the preferential order to relapse over hospitalisation, and assessed hospitalisation risk and relapse risk individually. Other secondary outcomes included all meta-analysable data, classed by relevance to effectiveness, efficacy, safety, quality of life, cognitive function, and other outcomes, and analysed by study design. Dichotomous outcomes were expressed as pooled RR and continuous outcomes as standardised mean difference (SMD). The protocol is registered with PROSPERO (CRD42019142094). Findings: We identified 14 687 records, of which 137 studies (397 319 patients) met the inclusion criteria (32 RCTs [23·4%; 8577 patients], 65 cohort studies [47·4%; 377 447 patients], and 40 pre–post studies [29·2%; 11 295 patients]) and were analysed. The quality of studies in terms of risk of bias varied across study designs and within each study design from low to high. LAIs were associated with a lower risk of hospitalisation or relapse than oral antipsychotics in each of the three study designs (RCTs: 29 studies, 7833 patients, RR 0·88 [95% CI 0·79–0·99], p=0·033; cohort studies: 44 studies, 106 136 patients, RR 0·92 [0·88–0·98], p=0·0044; pre–post studies: 28 studies, 17 876 patients, RR 0·44 [0·39–0·51], p<0·0001). This association was maintained across the study designs when we reversed the preferential order to risk of relapse over hospitalisation, and in individual analysis of hospitalisation risk. The association was maintained only in pre–post studies for relapse risk alone. In all other outcomes related to effectiveness, efficacy, safety, quality of life, cognitive function, and other outcomes, LAIs were more beneficial than oral antipsychotics in 60 (18·3%) of 328 comparisons, not different in 252 (76·8%) comparisons, and less beneficial in 16 (4·9%) comparisons when analysed by study design. Significant heterogeneity was observed across all three study designs. Publication biases were apparent in cohort and pre-post studies, but effect sizes were similar after trim-and-fill analyses. Interpretation: Although study designs have strengths and weaknesses, including potential low quality of observational studies, we consistently identified significant benefit with LAIs versus oral antipsychotics in preventing hospitalisation or relapse, in settings ranging from restricted research (RCTs) to real-word application (cohort and pre–post studies). Our findings suggest that increased clinical use of LAIs could improve outcomes in schizophrenia. Funding: None. Translations: For the Chinese, French, German, Italian, Japanese, Portugese and Spanish translations of the abstract see Supplementary Materials section.

Background: The COVID-19 pandemic has altered daily routines and family functioning, led to closing schools, and dramatically limited social interactions worldwide. Measuring its impact on mental health of vulnerable children and adolescents is crucial. Methods: The Collaborative Outcomes study on Health and Functioning during Infection Times (COH-FIT – www.coh-fit.com) is an on-line anonymous survey, available in 30 languages, involving >230 investigators from 49 countries supported by national/international professional associations. COH-FIT has thee waves (until the pandemic is declared over by the WHO, and 6–18 months plus 24–36 months after its end). In addition to adults, COH-FIT also includes adolescents (age 14–17 years), and children (age 6–13 years), recruited via non-probability/snowball and representative sampling and assessed via self-rating and parental rating. Non-modifiable/modifiable risk factors/treatment targets to inform prevention/intervention programs to promote health and prevent mental and physical illness in children and adolescents will be generated by COH-FIT. Co-primary outcomes are changes in well-being (WHO-5) and a composite psychopathology P-Score. Multiple behavioral, family, coping strategy and service utilization factors are also assessed, including functioning and quality of life. Results: Up to June 2021, over 13,000 children and adolescents from 59 countries have participated in the COH-FIT project, with representative samples from eleven countries. Limitations: Cross-sectional and anonymous design. Conclusions: Evidence generated by COH-FIT will provide an international estimate of the COVID-19 effect on children's, adolescents’ and families’, mental and physical health, well-being, functioning and quality of life, informing the formulation of present and future evidence-based interventions and policies to minimize adverse effects of the present and future pandemics on youth.

The importance of workers' well-being has been recognized in recent years. The assessment of well-being has been subjective, and few studies have sought potential biomarkers of well-being to date. This study examined the relationship between well-being and the LF/HF ratio, an index of heart rate variability that reflects sympathetic and parasympathetic nerve activity. Pulse waves were measured using photoplethysmography through a web camera attached to the computer used by each participant. The participants were asked to measure their pulse waves while working for 4 weeks, and well-being was assessed using self-reported measures such as the Satisfaction With Life Scale (SWLS), the Positive and Negative Affect Schedule (PANAS), and the Flourishing Scale (FS). Each of the well-being scores were split into two groups according to the median value, and the LF/HF ratio during work, as well as the number of times an LF/HF ratio threshold was either exceeded or subceeded, were compared between the high and low SWLS, positive emotion, negative emotion, and FS groups. Furthermore, to examine the effects of the LF/HF ratio and demographic characteristics on well-being, a multiple regression analysis was conducted. Data were obtained from 169 participants. The results showed that the low FS group had a higher mean LF/HF ratio during work than the high FS group. No significant differences were seen between the high and low SWLS groups, the high and low positive emotion groups, or the high and low negative emotion groups. The multiple regression analysis showed that the mean LF/HF ratio during work affected the FS and SWLS scores, and the number of times the mean LF/HF ratio exceeded +3 SD had an effect on the positive emotion. No effect of the LF/HF ratio on negative emotions was shown. The LF/HF ratio might be applicable as an objective measure of well-being.

Background Growing attention is paid to the association between alterations in the gut microbiota and their metabolites in patients with psychiatric disorders. Our study aimed to determine how gut microbiota and metabolomes are related to the sleep quality among patients with depression and anxiety disorders by analyzing the datasets of our previous study. Methods Samples were collected from 40 patients (depression: 32 patients [80.0%]); anxiety disorders: 8 patients [20.0%]) in this study. Gut microbiomes were analyzed using 16S rRNA gene sequencing and gut metabolomes were analyzed by a mass spectrometry approach. Based on the Pittsburgh Sleep Quality Index (PSQI), patients were categorized into two groups: the insomnia group (PSQI score ≥ 9, n = 20) and the non-insomnia group (PSQI score < 9, n = 20). Results The insomnia group showed a lower alpha diversity in the Chao1 and Shannon indices than the non-insomnia group after the false discovery rate (FDR) correction. The relative abundance of genus Bacteroides showed a positive correlation with PSQI scores in the non-insomnia group. The concentrations of glucosamine and N-methylglutamate were significantly higher in the insomnia group than in the non-insomnia group. Conclusions Our findings suggest that specific taxa could affect the sleep quality among patients with depression and anxiety disorders. Further studies are needed to elucidate the impact of sleep on specific gut microbiota and metabolomes in depression and anxiety disorders.

Japan, the world's most rapidly aging society, faces increasing financial strains related to personalized dementia care. The government has shifted its focus from prevention to coexistence with dementia, as outlined in the 2023 Basic Act on Dementia. Emphasis on financial inclusion aligns with the G20's 2019 "Fukuoka Policy Priorities on Aging and Financial Inclusion", which addresses financial exclusion due to cognitive decline and poor financial literacy. While economic activity among older adults is already hampered by legal challenges and risks associated with dementia, outcomes are expected to worsen as the assets of older adults with dementia are projected to reach 215 trillion JPY ($1.4 trillion USD) by 2030. Government measures and research in financial gerontology advocate for protecting older adults and promoting flexible financial practices. Enhanced efforts and shared research outcomes are crucial for Japan to be a leader as an advanced aging society.

International studies measuring wellbeing/multidimensional mental health before/ during the COVID-19 pandemic, including representative samples for >2 years, identifying risk groups and coping strategies are lacking. COH-FIT is an online, international, anonymous survey measuring changes in well-being (WHO-5) and a composite psychopathology P-score, and their associations with COVID-19 deaths/restrictions, 12 a-priori defined risk individual/cumulative factors, and coping strategies during COVID-19 pandemic (26/04/2020-26/06/2022) in 30 languages (representative, weighted non-representative, adults). T-test, χ2, penalized cubic splines, linear regression, correlation analyses were conducted. Analyzing 121,066/142,364 initiated surveys, WHO-5/P-score worsened intra-pandemic by 11.1±21.1/13.2±17.9 points (effect size d=0.50/0.60) (comparable results in representative/weighted non-probability samples). Persons with WHO-5 scores indicative of depression screening (<50, 13% to 32%) and major depression (<29, 3% to 12%) significantly increased. WHO-5 worsened from those with mental disorders, female sex, COVID-19-related loss, low-income country location, physical disorders, healthcare worker occupations, large city location, COVID-19 infection, unemployment, first-generation immigration, to age=18-29 with a cumulative effect. Similar findings emerged for P-score. Changes were significantly but minimally related to COVID-19 deaths, returning to near-pre-pandemic values after >2 years. The most subjectively effective coping strategies were exercise and walking, internet use, social contacts. Identified risk groups, coping strategies and outcome trajectories can inform global public health strategies.

Background: Asking questions is common in conversations, and while asking questions, we need to listen carefully to what others say and consider the perspective our questions adopt. However, difficulties persist in verifying the effect of asking questions on older adults’ cognitive function due to the lack of a standardized system for conducting experiments at participants’ homes. Objective: This study examined the intervention effect of cognitive training moderated by robots on healthy older adults. A focus on the feasibility of the intervention at participants’ homes was also maintained. Feasibility was evaluated by considering both the dropout rate during the intervention and the number of questions posed to each participant during the experiment. Methods: We conducted a randomized controlled trial with 81 adults older than 65 years. Participants were recruited through postal invitations and then randomized into 2 groups. The intervention group (n=40) received sessions where participants listened to photo-integrated stories and posed questions to the robots. The control group (n=41) received sessions where participants listened to photo-integrated stories and only thanked the robots for confirming participation. The participants participated in 12 dialogue sessions for 2-3 weeks. Scores of global cognitive functioning tests, recall tests, and verbal fluency tasks measured before and after the intervention were compared between the 2 groups. Results: There was no significant intervention effect on the Telephone Interview for Cognitive Status-Japanese scores, recall tests, and verbal fluency tasks. Additionally, our study successfully concluded with no participant dropouts at follow-up, confirming the feasibility of our approach. Conclusions: There was no statistically significant evidence indicating intervention benefits for cognitive functioning. Although the feasibility of home-based interventions was demonstrated, we identified areas for improvement in the future, such as setting up more efficient session themes. Further research is required to identify the effectiveness of an improved cognitive intervention involving the act of asking questions.

The development of a cost-effective digital biomarker for detecting dementia is highly needed. While numerous studies have explored dementia detection through speech and natural language analysis, only a few studies have focused on dementia detection using face video recordings, and more in-depth research is needed. In this paper, we propose a method for detecting dementia and mild cognitive impairment (MCI), a pre-dementia stage, by utilizing four types of facial expression features extracted from recorded videos of participants. These features include Action Units, emotion categories, Valence-Arousal, and face embeddings. From the above features obtained from each video frame, various statistical information was extracted and used as features, and predictions were performed using a decision tree-based model. Our method was evaluated using face video recordings during conversations. The method achieved an area under the receiver operating characteristic curve (AUC) of 0.933 for dementia detection and 0.889 for MCI detection. Statistical analysis of facial expression features revealed that participants with dementia had fewer positive emotions, more negative emotions, and lower valence and arousal than healthy participants. These results indicate that the proposed method could serve as an explainable screening tool for the early detection of dementia and MCI.

Alzheimer's disease (AD) stands as a prevalent form of dementia, primarily impacting the elderly, and presently lacking any known cure. Early screening and intervention have demonstrated efficacy for both current and prospective treatments. With the global population aging, early detection of AD has become imperative. Language-based approaches for dementia detection have received considerable attention in research due to dementia's nature of hurting cognitive and language functions. Facial features, on the other hand, have been comparatively neglected despite their potential in identifying AD, due to a lack of usable dataset. AD patients show different patterns while displaying facial expressions because of Apathy, a phenomenon common in AD patients. This makes facial features non-negligible in dementia detection. The face mesh, comprising 478 landmarks, serves as a rich and informative facial representation. This study concentrates on leveraging face mesh data for AD detection. We explored data augmentation techniques to address the challenges of over-fitting, particularly impactful in training neural networks on a small dataset. Deep learning methods, including transformer, Convolutional Neural Networks (CNN), and Long Short-Term Memory (LSTM) networks, were employed to classify dementia patients and healthy subjects. By augmenting the dataset by rotation, shift, and scaling, we improved the performance of the deep learning methods. The combined use of 1D CNN networks and data augmentation method reported the highest accuracy of 76%. These results confirmed the effectiveness of our data augmentation methods. They also underscore the potential of face-related features in enhancing early detection approaches for AD.

© 2020 American Medical Association. All rights reserved. Background: During the COVID-19 pandemic, the use of telemedicine as a way to reduce COVID-19 infections was noted and consequently deregulated. However, the degree of telemedicine regulation varies from country to country, which may alter the widespread use of telemediine. This study aimed to clarify the telepsychiatry regulations for each collaborating country/region before and during the COVID-19 pandemic. Methods: We used snowball sampling within a global network of international telepsychiatry experts. Thirty collaborators from 17 different countries/regions responded to a questionnaire on barriers to the use and implementation of telepsychiatric care, including policy factors such as regulations and reimbursement at the end of 2019 and as of May 2020. Results: Thirteen of 17 regions reported a relaxation of regulations due to the pandemic; consequently, all regions surveyed stated that telepsychiatry was now possible within their public healthcare systems. In some regions, restrictions on prescription medications allowed via telepsychiatry were eased, but in 11 of the 17 regions, there were still restrictions on prescribing medications via telepsychiatry. Lower insurance reimbursement amounts for telepsychiatry consultations versus in-person consultations were reevaluated in 4 regions, and consequently, in 15 regions telepsychiatry services were reimbursed at the same rate (or higher) than in-person consultations during the COVID-19 pandemic. Conclusions: Our results confirm that, due to COVID-19, the majority of countries surveyed are altering telemedicine regulations that had previously restricted the spread of telemedicine. These findings provide information that could guide future policy and regulatory decisions, that facilitate greater scale and spread of telepsychiatry globally.

© 2021 Elsevier Inc. Background: Electroconvulsive therapy (ECT) is the most effective treatment for severe depression. Recent neuroimaging studies have consistently reported that ECT induces volume increases in widely distributed brain regions. However, it still remains unclear about ECT-induced volume changes associated with clinical improvement. Methods: Longitudinal assessments of structural magnetic resonance imaging were conducted in 48 participants. Twenty-seven elderly melancholic depressed individuals (mean 67.5 ± 8.1 years old; 19 female) were scanned before (TP1) and after (TP2) ECT. Twenty-one healthy controls were also scanned twice. Whole-brain gray matter volume (GMV) was analyzed via group (remitters, nonremitters, and controls) by time (TP1 and TP2) analysis of covariance to identify ECT-related GMV changes and GMV changes specific to remitters. Within-subject and between-subjects correlation analyses were conducted to investigate the associations between clinical improvement and GMV changes. Depressive symptoms were evaluated using the 17-item Hamilton Depression Rating Scale (HAM-D), and remission was defined as HAM-D total score ≤ 7. Results: Bilateral ECT increased GMV in multiple brain regions bilaterally regardless of clinical improvement. Remitters showed a larger GMV increase in the right-lateralized frontolimbic brain regions compared to nonremitters and healthy controls. GMV changes in the right hippocampus/amygdala and right middle frontal gyrus showed correlations with clinical improvement in within−/between-subjects correlation analyses. Conclusions: ECT-induced GMV increase in the right frontolimbic regions was associated with clinical remission.

Background There are limited data about the association between body mass index (BMI), glycemic variability (GV), and life-related factors in healthy nondiabetic adults. Methods This cross-sectional study was carried out within our ethics committee-approved study called “Exploring the impact of nutrition advice on blood sugar and psychological status using continuous glucose monitoring (CGM) and wearable devices”. Prediabetes was defined by the HbA1c level of 5.7–6.4% and /or fasting glucose level of 100–125 mg/dL. Glucose levels and daily steps were measured for 40 participants using Free Style Libre and Fitbit Inspire 2 under normal conditions for 14 days. Dietary intakes and eating behaviors were assessed using a brief-type self-administered dietary history questionnaire and a modified questionnaire from the Obesity Guidelines. Results All indices of GV were higher in the prediabetes group than in the healthy group, but a significant difference was observed only in mean amplitude of glycemic excursions (MAGE). In the multivariate analysis, only the presence of prediabetes showed a significant association with the risk of higher than median MAGE (Odds, 6.786; 95% CI, 1.596–28.858; P = 0.010). Additionally, the underweight (BMI < 18.5) group had significantly higher value in standard deviation (23.7 ± 3.5 vs 19.8 ± 3.7 mg/dL, P = 0.038) and coefficient variability (22.6 ± 4.6 vs 18.4 ± 3.2%, P = 0.015), compared to the normal group. This GV can be partially attributed to irregularity of eating habits. On the contrary, the overweight (BMI > 25) group had the longest time above the 140 or 180 mg/dL range, which may be due to eating style and taking fewer steps (6394 ± 2337 vs 9749 ± 2408 steps, P = 0.013). Conclusions Concurrent CGM with diet and activity monitoring could reduce postprandial hyperglycemia through assessment of diet and daily activity, especially in non- normal weight individuals.

We report a patient with logopenic variant primary progressive aphasia (lv-PPA) who was diagnosed as having non-Alzheimer's disease (AD) tauopathy after multiple biophysical/biological examinations, including amyloid and 18F-florzolotau tau positron emission tomography (PET), had been performed. A woman in her late 60s who had previously been diagnosed as having AD was referred to us for a further, detailed examination. She had been unaware of any symptoms at the time of AD diagnosis, but she subsequently became gradually aware of a speech impairment. She talked nearly completely and fluently, although she occasionally exhibited word-finding difficulty and made phonological errors during naming, word fluency testing, and sentence repetition; these findings met the criteria for the diagnosis of lv-PPA, which is known to be observed more commonly in AD than in other proteinopathies. Magnetic resonance imaging, single photon emission computed tomography, and plasma phosphorylated tau and plasma neurofilament light chain measurements showed an AD-like pattern. However, both 11C-Pittsburgh compound-B and 18F-florbetaben amyloid PET showed negative results, whereas 18F-florzolotau tau PET yielded positive results, with radio signals predominantly in the left superior temporal gyrus, middle temporal gyrus, supramarginal gyrus, and frontal operculum. Whole-genome sequencing revealed no known dominantly inherited mutations in AD or frontotemporal lobar degeneration genes, including the genes encoding amyloid precursor protein, microtubule-associated protein tau, presenilin 1 and 2. To the best of our knowledge, this patient was a rare case of lv-PPA who was diagnosed as having non-AD tauopathy based on the results of multiple examinations, including whole-genome sequencing, plasma measurement, and amyloid and 18F-florzolotau tau PET. This case underscores the clinicopathologically heterogeneous nature of this syndrome.