Min-Wei Huang

The high rate of repeat attempts among individuals who have previously attempted suicide presents a critical challenge in public health and suicide prevention. While early and targeted intervention is crucial for this high-risk group, effectively identifying those most likely to re-attempt is a persistent difficulty, especially when public health resources are limited. This creates a pressing need for accurate and practical risk assessment tools. This study aims to address this gap by using machine learning to analyze a nationwide suicide surveillance database to identify key predictors of repeat suicide attempts and develop a robust predictive model to aid in resource allocation and early intervention. This study analyzed data from 32,701 individuals, encompassing 31 features, recorded in Taiwan's National Suicide Surveillance System for 2020. We employed binary decision tree regression with multiple feature selection techniques to identify significant predictors of recurrent suicide attempts. A prediction model was then developed without requiring biological samples. History of mental illness, specific age groups, and supervision status of mentally ill patients emerged as primary influences on repeated suicide attempts. The prediction model achieved 66.3% accuracy in identifying potential repeat attempters, with a 57.9% success rate for predicting re-suicide events. These findings align with recent literature on suicide recurrence risk factors. This study provides a practical risk assessment tool that enables early intervention for high-risk individuals without invasive biological sampling. The insights offer valuable guidance for government suicide prevention policies, especially when prioritizing cases with limited resources. Future enhancements through broader data integration and interdisciplinary cooperation could establish a more comprehensive prevention system to reduce the societal impact of suicidal behavior. Not applicable as this was a retrospective database analysis rather than a randomized controlled trial.

Anorexia nervosa (AN) is a serious and potentially life-threatening eating disorder characterized by starvation and malnutrition, a high prevalence of coexisting psychiatric conditions, marked treatment resistance, frequent medical complications, and a substantial risk of death. Body mass index (BMI) is a key measure of treatment outcome of AN and it is necessary to evaluate the long-term prognosis of AN. This study aimed to better assess the BMI course trend between different medications and timepoints in order to improve AN treatment in clinical practice. During the period 2010–2021, we retrospectively reviewed historical data of all patients diagnosed with AN. There were two groups in this study, which were based on the duration of follow-up. Group A was a 6-month follow-up group, comprising 93 patients (mean age 19.6 ± 6.8 years), with BMI assessed at three consecutive time points: first outpatient visit (T0), three months follow-up (T3), and six months follow-up (T6). Group B was a 12-month follow-up group comprising 36 patients (mean age 17.0 ± 5.2 years) with BMI assessed at five consecutive time points: first outpatient visit (T0), three months follow-up (T3), six months follow-up (T6), nine months follow-up (T9), and twelve months follow-up (T12). In our study, we retrospectively compared BMI courses based on patients’ usage of medication using the following variables: single medication, switching medications, combined medications, and without medications. The primary outcome measurement was BMI recorded at the 6-month follow-up and the 12-month follow-up respectively. In our study, which was conducted at Taichung Veterans General Hospital, we reviewed outpatient medical records of all patients with AN who were seen at the hospital during the period 2010–2021. In Group A (6-month follow-up), patients treated with antidepressants showed a mean BMI increase of 1.3 (p < 0.001); patients treated with antipsychotics showed a mean BMI increase of 1.1 (p = 0.01); patients treated with switching medications showed a mean BMI increase of 0.1 (p = 0.397); patients treated with combined medications showed a mean BMI increase of 0.5 (p = 0.208); and patients treated without medications showed a mean BMI increase of 0.1 (p = 0.821). The results indicated that patients with AN had a significant BMI increase after treatment with antidepressants and antipsychotics in the 6-month follow-up group. In Group B (12-month follow-up), patients treated with antidepressants showed a mean BMI increase of 2.7 (p < 0.001); patients treated with antipsychotics showed a mean BMI increase of 2.8 (p = 0.168); patients treated with switching medications showed a mean BMI decrease of 0.8 (p = 0.595); patients treated with combined medications showed a mean BMI increase of 1.6 (p = 0.368); and patients treated without medications showed a mean BMI increase of 1.0 (p = 0.262). The results indicated that patients with AN had a significant BMI increase after treatment with antidepressants at the 12-month follow-up. AN is a complex disease caused by multiple factors. Evaluating its long-term prognosis is crucial. Our study provides insights and highlights three key findings: 1) medication adherence is crucial in treating AN, 2) frequent switching of medications may not promote weight gain and may also require a re-establishment of rapport with patients with AN, and 3) pharmacotherapy, especially antidepressants, is more effective than no treatment. Further research is needed to confirm these findings.

Breast cancer prediction datasets are usually class imbalanced, where the number of data samples in the malignant and benign patient classes are significantly different. Over-sampling techniques can be used to re-balance the datasets to construct more effective prediction models. Moreover, some related studies have considered feature selection to remove irrelevant features from the datasets for further performance improvement. However, since the order of combining feature selection and over-sampling can result in different training sets to construct the prediction model, it is unknown which order performs better. In this paper, the information gain (IG) and genetic algorithm (GA) feature selection methods and the synthetic minority over-sampling technique (SMOTE) are used for different combinations. The experimental results based on two breast cancer datasets show that the combination of feature selection and over-sampling outperform the single usage of either feature selection and over-sampling for the highly class imbalanced datasets. In particular, performing IG first and SMOTE second is the better choice. For other datasets with a small class imbalance ratio and a smaller number of features, performing SMOTE is enough to construct an effective prediction model.

For decades, scientists have been trying to solve the problem of dementia, with no cure currently available. Semantic–lexical impairment is well established as the early critical sign of dementia, although there are still gaps in knowledge that must be investigated. In this study, we used fMRI to observe the neural activity of 31 subjects, including 16 HC (Healthy Control) and 15 AD (Alzheimer’s Disease), who participated in the naming task. The neuropsychology profile of HC (Healthy Control) and AD (Alzheimer’s Disease) are discussed in this study. The involvement of FG (Fusiform Gyrus) and IFG (Inferior Frontal Gyrus) shows dominant activation in both of the groups. We observed a decrease in neural activity in the AD group, resulting in semantic deficit problems in this preliminary study. Furthermore, ROI analysis was performed and revealed both hyperactivation and hypoactivation in the AD group. The compensatory mechanism demonstrated during the task, due to the effort required to identify an animal’s name, represents the character profile of AD.

The main purpose of this research is to evaluate the differences in the reaction time and accuracy rate of three categories of subjects using our serious games. Thirty-seven subjects were divided into three groups: normal (n₁ = 16), MCI (Mild Cognitive Impairment) (n₂ = 10), and dementia—moderate-to-severe (n₃ = 11) groups based on the MMSE (Mini Mental State Examination). Two serious games were designed: (1) whack-a-mole and (2) hit-the-ball. Two dependent variables, reaction time and accuracy rate, were statistically analyzed to compare elders’ performances in the games among the three groups for three levels of speed: slow, medium, and fast. There were significance differences between the normal group, the MCI group, and the moderate-to-severe dementia group in both the reaction-time and accuracy-rate analyses. We determined that the reaction times of the MCI and dementia groups were shorter compared to those of the normal group, with poorer results also observed in accuracy rate. Therefore, we conclude that our serious games have the feasibility to evaluate reaction performance and could be used in the daily lives of elders followed by clinical treatment in the future.

In this study, the international affective picture system was used to evoke emotion, and then the corresponding signals were collected. The features from different points of brainwaves, frequency, and entropy were used to identify normal, moderately, and markedly ill schizophrenic patients. The signals were collected and preprocessed. Then, the signals were separated according to three types of emotions and five frequency bands. Finally, the features were calculated using three different methods of entropy. For classification, the features were divided into different sections and classification using support vector machine (principal components analysis on 95%). Finally, simple regression and correlation analysis between the total scores of positive and negative syndrome scale and features were used. At first, we observed that to classify normal and markedly ill schizophrenic patients, the identification result was as high as 81.5%, and therefore, we further explored moderately and markedly ill schizophrenic patients. Second, the identification rate in both moderately and markedly ill schizophrenic patient was as high as 79.5%, which at the Fz point signal in high valence low arousal fragments was calculated using the ApEn methods. Finally, the total scores of positive and negative syndrome scale were used to analyze the correlation with the features that were the five frequency bands at the Fz point signal. The results show that the p value was less than .001 at the beta wave in the 15–18 Hz frequency range.

Schizophrenia is a neurological disease characterized by alterations to patients’ cognitive functions and emotional expressions. Relevant studies often use magnetic resonance imaging (MRI) of the brain to explore structural differences and responsiveness within brain regions. However, as this technique is expensive and commonly induces claustrophobia, it is frequently refused by patients. Thus, this study used non-contact infrared thermal facial images (ITFIs) to analyze facial temperature changes evoked by different emotions in moderately and markedly ill schizophrenia patients. Schizophrenia is an emotion-related disorder, and images eliciting different types of emotions were selected from the international affective picture system (IAPS) and presented to subjects during ITFI collection. ITFIs were aligned using affine registration, and the changes induced by small irregular head movements were corrected. The average temperatures from the forehead, nose, mouth, left cheek, and right cheek were calculated, and continuous temperature changes were used as features. After performing dimensionality reduction and noise removal using the component analysis method, multivariate analysis of variance and the Support Vector Machine (SVM) classification algorithm were used to identify moderately and markedly ill schizophrenia patients. Analysis of five facial areas indicated significant temperature changes in the forehead and nose upon exposure to various emotional stimuli and in the right cheek upon evocation of high valence low arousal (HVLA) stimuli. The most significant P-value (lower than 0.001) was obtained in the forehead area upon evocation of disgust. Finally, when the features of forehead temperature changes in response to low valence high arousal (LVHA) were reduced to 9 using dimensionality reduction and noise removal, the identification rate was as high as 94.3%. Our results show that features obtained in the forehead, nose, and right cheek significantly differed between moderately and markedly ill schizophrenia patients. We then chose the features that most effectively distinguish between moderately and markedly ill schizophrenia patients using the SVM. These results demonstrate that the ITFI analysis protocol proposed in this study can effectively provide reference information regarding the phase of the disease in patients with schizophrenia.

Despite an increasing number of reports on the associations between chronic occupational stress and structural and functional changes of the brain, the underlying neural correlates of perceived occupational stress is still not clear. Perceived stress reflects the extents to which situations are appraised as stressful at a given point in one’s life. Using near-infrared spectroscopy, we investigated the associations between perceived occupational stress and cortical activity over the bilateral frontotemporal regions during a verbal fluency test. Sixty-eight participants (17 men, 51 women), 20–62 years of age were recruited. Perceived occupational stress was measured using the Chinese version of Job Content Questionnaire, and the Chinese version of the Copenhagen Burnout Inventory. We found statistically significant negative associations between occupational burnout and brain cortical activity over the fronto-polar and dorsolateral prefrontal cortex during the VFT (r = −0.343 to −0.464). In conclusion, our research demonstrated a possible neural basis of perceived occupational stress that are distributed across the prefrontal cortex.

We investigated whether variation in the dopamine D2 receptor gene (DRD2) and tri-dimensional personality questionnaire (TPQ) scores could be used to aid adjustment of daily methadone requirements of heroin addicts. DRD2 TaqI B polymorphisms and TPQ scores were determined in 138 male Taiwanese heroin addicts who were receiving methadone treatment. Borderline index (harm avoidance + novelty seeking—reward dependence) was calculated for each subject, and three groups were defined: high (mean from all subjects plus 1 standard deviation, or greater), low (half of the calculated high score, or lower) and medium (all values between the high and low scores). No significant differences in age (p = 0.60), mean methadone dose (p = 0.75) or borderline index group (p = 0.25) were observed between subjects bearing the B1/B1, B1/B2 and B2/B2 DRD2 TaqI genotypes. Among the individuals with low (≤10), medium (11–20) and high (≥21) borderline index scores, there was a significant difference in mean methadone dose (p = 0.04), but not age (p = 0.90). Further analysis showed that mean methadone dose was significantly higher in subjects with low borderline index scores than in those with high scores (62.5 vs. 47.0 mg/day, p = 0.03). The odds ratio for a daily methadone requirement ≥60 mg (median dose across the 138 subjects) was 2.64-fold greater in the low borderline index group than in the high group (p = 0.04). Although the DRD2 TaqI B genotype was not associated with methadone use requirements, borderline index was revealed as a potential predictive marker for the adjustment of methadone dosage requirements in heroin addicts.