Solaphat Hemrungrojn
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Solaphat Hemrungrojn has a total of 43 co-authors. Here are the top co-authors by number of shared publications:
- Michael Maes
Department of Psychiatry, Faculty of Medicine, Chulalongkorn University, Bangkok, Thailand
6 shared publications - Arisara Amrapala
Department of Psychiatry, Faculty of Medicine, Chulalongkorn University, Bangkok, Thailand
3 shared publications Kulvara Lapanan
Neuroscience Research Australia, Randwick, New South Wales, Australia; School of Population Health, Faculty of Medicine and Health, University of New South Wales, Sydney, New South Wales, Australia
3 shared publicationsArisara Amrapala
Center of Excellence in Digital and AI for Mental Health, Faculty of Engineering, Chulalongkorn University, Bangkok, Thailand; Cognitive Fitness and Biopsychiatry Technology Research Unit, Faculty of Medicine, Chulalongkorn University, Bangkok, Thailand; Department of Psychiatry, Faculty of Medicine, Chulalongkorn University, Bangkok, Thailand
2 shared publications- Chaipat Chunharas
Cognitive Clinical and Computational Neuroscience Center of Excellence, Department of Medicine, King Chulalongkorn Memorial Hospital, Chulalongkorn University
2 shared publications Chanyanart Kiattiporn-Opas
Center of Excellence in Digital and AI for Mental Health, Faculty of Engineering, Chulalongkorn University, Bangkok, Thailand
2 shared publications
Latest Publications
- Open Access
Circulating Apolipoprotein Profiles and Their Associations with Mild Cognitive and Behavioral Impairment in Older Adults: A Cross-Sectional Study
Background: Mild cognitive impairment (MCI) and mild behavioral impairment (MBI) are overlapping prodromal states associated with dementia risk. Circulating apolipoproteins (Apos) may be linked to cognitive and behavioral changes, but their relationships remain unclear. Objective: Examine plasma concentrations of multiple Apos in prodromal dementia. Methods: This cross-sectional study included 148 Thai participants aged 60-75 years: 56 cognitively normal controls, 47 with MCI without MBI, and 45 with MBI with or without MCI. We measured ApoA1, ApoA2, ApoB, ApoC1, ApoC3, ApoD, ApoE, ApoH, and ApoJ, five lipids, and APOE genotype, and computed an atherogenic Apo composite (ApoB-ApoC1-ApoC3). Primary analyses used multivariable linear regression of quantitative cognitive and behavioral outcomes. Group comparisons were secondary and served to contextualize the dimensional regression findings. Results: Across Montreal Cognitive Assessment (MoCA) models, 20.7-28.2% of variance was explained by combinations of ApoH, ApoC3, APOE ε4 carrier status, age, and male sex (inverse associations), and free cholesterol and education (positive associations). The atherogenic composite was inversely associated with MoCA total, memory index, and delayed recall scores, and with Consortium to Establish a Registry for Alzheimer’s Disease (CERAD) verbal fluency and word list recall. No significant predictors emerged for Neuropsychiatric Inventory (NPI) scores or NPI caregiver distress. All nine Apos differed significantly between groups, with levels highest in MCI, intermediate in MBI, and lowest in controls. Conclusions: Circulating Apo profiles differed across controls, MCI, and MBI groups and were more consistently associated with cognitive than behavioral outcomes. ApoH and the atherogenic Apo composite warrant further investigation as candidate biomarkers of early cognitive dysfunction.
Open AccessCirculating Apolipoprotein Profiles and Their Associations with Mild Cognitive and Behavioral Impairment in Older Adults: A Cross-Sectional Study
Background: Mild cognitive impairment (MCI) and mild behavioral impairment (MBI) are overlapping prodromal states associated with dementia risk. Circulating apolipoproteins (Apos) may be linked to cognitive and behavioral changes, but their relationships remain unclear. Objective: Examine plasma concentrations of multiple Apos in prodromal dementia. Methods: This cross-sectional study included 148 Thai participants aged 60-75 years: 56 cognitively normal controls, 47 with MCI without MBI, and 45 with MBI with or without MCI. We measured ApoA1, ApoA2, ApoB, ApoC1, ApoC3, ApoD, ApoE, ApoH, and ApoJ, five lipids, and APOE genotype, and computed an atherogenic Apo composite (ApoB-ApoC1-ApoC3). Primary analyses used multivariable linear regression of quantitative cognitive and behavioral outcomes. Group comparisons were secondary and served to contextualize the dimensional regression findings. Results: Across Montreal Cognitive Assessment (MoCA) models, 20.7-28.2% of variance was explained by combinations of ApoH, ApoC3, APOE ε4 carrier status, age, and male sex (inverse associations), and free cholesterol and education (positive associations). The atherogenic composite was inversely associated with MoCA total, memory index, and delayed recall scores, and with Consortium to Establish a Registry for Alzheimer’s Disease (CERAD) verbal fluency and word list recall. No significant predictors emerged for Neuropsychiatric Inventory (NPI) scores or NPI caregiver distress. All nine Apos differed significantly between groups, with levels highest in MCI, intermediate in MBI, and lowest in controls. Conclusions: Circulating Apo profiles differed across controls, MCI, and MBI groups and were more consistently associated with cognitive than behavioral outcomes. ApoH and the atherogenic Apo composite warrant further investigation as candidate biomarkers of early cognitive dysfunction.
- Open Access
- 1 September 2024
- Published by Faculty of Medicine Siriraj Hospital, Mahidol University in Siriraj Medical Journal
- Vol. 76 (9) , 620-629
Development and Evaluation of the DMIND Questionnaire: Preparing for AI Integration into an Effective Depression Screening Tool
Objective: Thailand’s mental health crisis is exacerbated by high demand and a shortage of mental health professionals. The research objective was to develop and validate the Detection and Monitoring Intelligence Network for Depression (DMIND) questionnaire, designed to be culturally relevant and easily administered in clinical settings. Crafted with expert input, items specifically conducive to artificial intelligence (AI) analysis were selected to facilitate the future development of an AI-assisted depression scoring model. This approach underscores the tool’s dual utility in both human-led and technology-enhanced diagnostics. Materials and Methods: We enrolled 81 participants from psychiatric and tertiary care hospitals in Bangkok. Participants were assessed using the DMIND questionnaire, followed by the Hamilton Depression Rating Scale (HDRS-17). Statistical analyses included the content validity index (CVI), Cronbach’s alpha, Pearson’s correlation coefficient, Cohen’s kappa, and receiver operating characteristic (ROC) analysis. The Liu method, Youden index, and nearest neighbor method were used to determine the optimal cut-off point Results: The DMIND questionnaire showed strong validity, with an item-level CVI (I-CVI) and scale-level CVI (S-CVI) exceeding 1.0, indicating strong consensus on its relevance and utility. The tool also demonstrated high internal consistency (Cronbach’s alpha = 0.96). ROC analysis showed an AUC of 0.88, indicating high accuracy in depression screening. An optimal cut-off score of 11.5 was identified, balancing predictive value and sensitivity. Conclusion: The DMIND questionnaire represents a significant advancement in innovative mental health diagnostics, addressing unmet clinical needs by providing accurate and efficient assessments capable of AI integration for further enhancing mental health service delivery in Thailand.
- 1 September 2024
- Published by Faculty of Medicine Siriraj Hospital, Mahidol University in Siriraj Medical Journal
- Vol. 76 (9) , 620-629
Open AccessDevelopment and Evaluation of the DMIND Questionnaire: Preparing for AI Integration into an Effective Depression Screening Tool
Objective: Thailand’s mental health crisis is exacerbated by high demand and a shortage of mental health professionals. The research objective was to develop and validate the Detection and Monitoring Intelligence Network for Depression (DMIND) questionnaire, designed to be culturally relevant and easily administered in clinical settings. Crafted with expert input, items specifically conducive to artificial intelligence (AI) analysis were selected to facilitate the future development of an AI-assisted depression scoring model. This approach underscores the tool’s dual utility in both human-led and technology-enhanced diagnostics. Materials and Methods: We enrolled 81 participants from psychiatric and tertiary care hospitals in Bangkok. Participants were assessed using the DMIND questionnaire, followed by the Hamilton Depression Rating Scale (HDRS-17). Statistical analyses included the content validity index (CVI), Cronbach’s alpha, Pearson’s correlation coefficient, Cohen’s kappa, and receiver operating characteristic (ROC) analysis. The Liu method, Youden index, and nearest neighbor method were used to determine the optimal cut-off point Results: The DMIND questionnaire showed strong validity, with an item-level CVI (I-CVI) and scale-level CVI (S-CVI) exceeding 1.0, indicating strong consensus on its relevance and utility. The tool also demonstrated high internal consistency (Cronbach’s alpha = 0.96). ROC analysis showed an AUC of 0.88, indicating high accuracy in depression screening. An optimal cut-off score of 11.5 was identified, balancing predictive value and sensitivity. Conclusion: The DMIND questionnaire represents a significant advancement in innovative mental health diagnostics, addressing unmet clinical needs by providing accurate and efficient assessments capable of AI integration for further enhancing mental health service delivery in Thailand.
- Open Access
Development and Evaluation of the DMIND Questionnaire: Preparing for AI Integration into an Effective Depression Screening Tool
Objective: Thailand’s mental health crisis is exacerbated by high demand and a shortage of mental health professionals. The research objective was to develop and validate the Detection and Monitoring Intelligence Network for Depression (DMIND) questionnaire, designed to be culturally relevant and easily administered in clinical settings. Crafted with expert input, items specifically conducive to artificial intelligence (AI) analysis were selected to facilitate the future development of an AI-assisted depression scoring model. This approach underscores the tool’s dual utility in both human-led and technology-enhanced diagnostics. Materials and Methods: We enrolled 81 participants from psychiatric and tertiary care hospitals in Bangkok. Participants were assessed using the DMIND questionnaire, followed by the Hamilton Depression Rating Scale (HDRS-17). Statistical analyses included the content validity index (CVI), Cronbach’s alpha, Pearson’s correlation coefficient, Cohen’s kappa, and receiver operating characteristic (ROC) analysis. The Liu method, Youden index, and nearest neighbor method were used to determine the optimal cut-off point Results: The DMIND questionnaire showed strong validity, with an item-level CVI (I- CVI) and scale-level CVI (S-CVI) exceeding 1.0, indicating strong consensus on its relevance and utility. The tool also demonstrated high internal consistency (Cronbach’s alpha = 0.96). ROC analysis showed an AUC of 0.88, indicating high accuracy in depression screening. An optimal cut-off score of 11.5 was identified, balancing predictive value and sensitivity. Conclusion: The DMIND questionnaire represents a significant advancement in innovative mental health diagnostics, addressing unmet clinical needs by providing accurate and efficient assessments capable of AI integration for further enhancing mental health service delivery in Thailand.
Open AccessDevelopment and Evaluation of the DMIND Questionnaire: Preparing for AI Integration into an Effective Depression Screening Tool
Objective: Thailand’s mental health crisis is exacerbated by high demand and a shortage of mental health professionals. The research objective was to develop and validate the Detection and Monitoring Intelligence Network for Depression (DMIND) questionnaire, designed to be culturally relevant and easily administered in clinical settings. Crafted with expert input, items specifically conducive to artificial intelligence (AI) analysis were selected to facilitate the future development of an AI-assisted depression scoring model. This approach underscores the tool’s dual utility in both human-led and technology-enhanced diagnostics. Materials and Methods: We enrolled 81 participants from psychiatric and tertiary care hospitals in Bangkok. Participants were assessed using the DMIND questionnaire, followed by the Hamilton Depression Rating Scale (HDRS-17). Statistical analyses included the content validity index (CVI), Cronbach’s alpha, Pearson’s correlation coefficient, Cohen’s kappa, and receiver operating characteristic (ROC) analysis. The Liu method, Youden index, and nearest neighbor method were used to determine the optimal cut-off point Results: The DMIND questionnaire showed strong validity, with an item-level CVI (I- CVI) and scale-level CVI (S-CVI) exceeding 1.0, indicating strong consensus on its relevance and utility. The tool also demonstrated high internal consistency (Cronbach’s alpha = 0.96). ROC analysis showed an AUC of 0.88, indicating high accuracy in depression screening. An optimal cut-off score of 11.5 was identified, balancing predictive value and sensitivity. Conclusion: The DMIND questionnaire represents a significant advancement in innovative mental health diagnostics, addressing unmet clinical needs by providing accurate and efficient assessments capable of AI integration for further enhancing mental health service delivery in Thailand.
- Open Access
Disorders in lipid metabolism, oxidative stress, and antioxidants in patients with amnestic mild cognitive impairment without major depression
Background: Amnestic mild cognitive impairment (aMCI) is characterized by changes in lipids and oxidative stress (OS). It is crucial to exclude patients with major depression (MDD) to accurately evaluate these biomarkers in aMCI. Aims: To examine lipid and oxidative stress biomarkers associated with aMCI versus normal controls. Methods: We performed a case-control analysis involving 61 individuals with aMCI (without MDD) and 60 healthy controls. We assessed the severity of aMCI, distress symptoms of old age, and lipid/OS biomarkers. Results: The levels of serum -SH groups were significantly higher in individuals with aMCI, while the levels of malondialdehyde (MDA) were significantly lower in the same group. Serum advanced oxidation protein products, glutathione, and folic acid did not show any notable variations. In individuals with aMCI, we observed an elevated apolipoprotein B (ApoB)/ApoA ratio, as well as decreased levels of high-density lipoprotein cholesterol (HDL), ApoA, and a reverse cholesterol transport (RCT) index. The simultaneous presence of aMCI and subclinical depressive symptoms is marked by elevated levels of triglycerides and ApoB, as well as decreased levels of ApoA and HDL. A significant portion of the variability (24.9%) in a quantitative MCI severity score can be attributed to -SH groups, age (positively), MDA and education (inversely). Discussion: The alterations in MDA and -SH levels in aMCI may potentially disrupt redox signaling, which can affect cell signaling and homeostatic setpoints. The interaction between aMCI and subclinical depressive symptoms can lead to increased atherogenicity and reduced antiatherogenic protection.
Open AccessDisorders in lipid metabolism, oxidative stress, and antioxidants in patients with amnestic mild cognitive impairment without major depression
Background: Amnestic mild cognitive impairment (aMCI) is characterized by changes in lipids and oxidative stress (OS). It is crucial to exclude patients with major depression (MDD) to accurately evaluate these biomarkers in aMCI. Aims: To examine lipid and oxidative stress biomarkers associated with aMCI versus normal controls. Methods: We performed a case-control analysis involving 61 individuals with aMCI (without MDD) and 60 healthy controls. We assessed the severity of aMCI, distress symptoms of old age, and lipid/OS biomarkers. Results: The levels of serum -SH groups were significantly higher in individuals with aMCI, while the levels of malondialdehyde (MDA) were significantly lower in the same group. Serum advanced oxidation protein products, glutathione, and folic acid did not show any notable variations. In individuals with aMCI, we observed an elevated apolipoprotein B (ApoB)/ApoA ratio, as well as decreased levels of high-density lipoprotein cholesterol (HDL), ApoA, and a reverse cholesterol transport (RCT) index. The simultaneous presence of aMCI and subclinical depressive symptoms is marked by elevated levels of triglycerides and ApoB, as well as decreased levels of ApoA and HDL. A significant portion of the variability (24.9%) in a quantitative MCI severity score can be attributed to -SH groups, age (positively), MDA and education (inversely). Discussion: The alterations in MDA and -SH levels in aMCI may potentially disrupt redox signaling, which can affect cell signaling and homeostatic setpoints. The interaction between aMCI and subclinical depressive symptoms can lead to increased atherogenicity and reduced antiatherogenic protection.
- Open Access
Distress symptoms of old age and mild cognitive impairment are two distinct dimensions in older adults without major depression: effects of adverse childhood experiences and negative life events
Background: Studies in old adults showed bidirectional interconnections between amnestic mild cognitive impairment (aMCI) and affective symptoms and that adverse childhood experiences (ACE) may affect both factors. Nevertheless, these associations may be confined to older adults with clinical depression. Aims: To delineate the relationship between clinical symptoms of aMCI and affective symptoms in older adults without major depression (MDD) or dysfunctions in activities of daily living (ADL). Methods: This case-control study recruited 61 participants with aMCI (diagnosed using Petersen’s criteria) and 59 older adults without aMCI and excluded subjects with MDD and ADL dysfunctions. Results: We uncovered 2 distinct dimensions, namely distress symptoms of old age (DSOA) comprising anxiety, depression, perceived stress and neuroticism scores, and mild cognitive dysfunctions (mCoDy) comprising episodic memory test scores, and the total Mini-Mental State Examination (MMSE) and Montreal Cognitive Assessment (MoCA) scores. A large part of the variance (37.9%) in DSOA scores was explained by ACE, negative life events (health and financial problems), a subjective feeling of cognitive decline, and education (all positively). While ACE and NLE have a highly significant impact on the DSOA, they are not associated with the mCoDy scores. Cluster analysis showed that the diagnosis of aMCI is overinclusive because some subjects with DSOA symptoms may be incorrectly classified as aMCI. Conclusions: The clinical impact is that clinicians should carefully screen older adults for DSOA after excluding MDD. DSOA might be misinterpreted as aMCI.
Open AccessDistress symptoms of old age and mild cognitive impairment are two distinct dimensions in older adults without major depression: effects of adverse childhood experiences and negative life events
Background: Studies in old adults showed bidirectional interconnections between amnestic mild cognitive impairment (aMCI) and affective symptoms and that adverse childhood experiences (ACE) may affect both factors. Nevertheless, these associations may be confined to older adults with clinical depression. Aims: To delineate the relationship between clinical symptoms of aMCI and affective symptoms in older adults without major depression (MDD) or dysfunctions in activities of daily living (ADL). Methods: This case-control study recruited 61 participants with aMCI (diagnosed using Petersen’s criteria) and 59 older adults without aMCI and excluded subjects with MDD and ADL dysfunctions. Results: We uncovered 2 distinct dimensions, namely distress symptoms of old age (DSOA) comprising anxiety, depression, perceived stress and neuroticism scores, and mild cognitive dysfunctions (mCoDy) comprising episodic memory test scores, and the total Mini-Mental State Examination (MMSE) and Montreal Cognitive Assessment (MoCA) scores. A large part of the variance (37.9%) in DSOA scores was explained by ACE, negative life events (health and financial problems), a subjective feeling of cognitive decline, and education (all positively). While ACE and NLE have a highly significant impact on the DSOA, they are not associated with the mCoDy scores. Cluster analysis showed that the diagnosis of aMCI is overinclusive because some subjects with DSOA symptoms may be incorrectly classified as aMCI. Conclusions: The clinical impact is that clinicians should carefully screen older adults for DSOA after excluding MDD. DSOA might be misinterpreted as aMCI.
- Open Access
Cerebral hemodynamics response to dual-task paradigms in older adults: A systematic review and meta-analysis
Background: Functional near-infrared spectroscopy (fNIRS) is a method to measure cerebral hemodynamics response. Changes in the prefrontal cortex during dual-tasking help to identify the risk of falling, particularly in older adults. Aims: To systematically review and meta-analyze the effects of dual-task paradigms on cerebral hemodynamics in older adults. Methods: The search was conducted in PubMed, Scopus, and Web of Science. A total of 22 studies comprising 1,841 older adults were included in the meta-analysis. Results: Inhibitory control and working memory tasks significantly increased HbO 2 in the PFC by 0.54 ( p < 0.01, 95%CI = 0.36 – 0.72) and 0.13 ( p < 0.01, 95%CI = 0.08 – 0.18) μmol/L, respectively. Overall, HbO 2 was significantly increased during dual-task paradigms by 0.36 μmol/L ( P < 0.01, 95%CI = 0.27 – 0.45). However, dual-task paradigms did not change HbR in the PFC ( P = 0.14, 95%CI = -0.05 – 0.01). Conclusion: Cognitive tasks related to inhibitory control required greater cognitive demands, indicating higher PFC activation during dual-task walking in older adults. This finding emphasizes the significance of assessing hemodynamic responses under dual-task paradigms to detect the risk of falling in older adults at an early stage.
Open AccessCerebral hemodynamics response to dual-task paradigms in older adults: A systematic review and meta-analysis
Background: Functional near-infrared spectroscopy (fNIRS) is a method to measure cerebral hemodynamics response. Changes in the prefrontal cortex during dual-tasking help to identify the risk of falling, particularly in older adults. Aims: To systematically review and meta-analyze the effects of dual-task paradigms on cerebral hemodynamics in older adults. Methods: The search was conducted in PubMed, Scopus, and Web of Science. A total of 22 studies comprising 1,841 older adults were included in the meta-analysis. Results: Inhibitory control and working memory tasks significantly increased HbO 2 in the PFC by 0.54 ( p < 0.01, 95%CI = 0.36 – 0.72) and 0.13 ( p < 0.01, 95%CI = 0.08 – 0.18) μmol/L, respectively. Overall, HbO 2 was significantly increased during dual-task paradigms by 0.36 μmol/L ( P < 0.01, 95%CI = 0.27 – 0.45). However, dual-task paradigms did not change HbR in the PFC ( P = 0.14, 95%CI = -0.05 – 0.01). Conclusion: Cognitive tasks related to inhibitory control required greater cognitive demands, indicating higher PFC activation during dual-task walking in older adults. This finding emphasizes the significance of assessing hemodynamic responses under dual-task paradigms to detect the risk of falling in older adults at an early stage.
- Open Access
Selecting the Most Important Features for Predicting Mild Cognitive Impairment from Thai Verbal Fluency Assessments
Mild cognitive impairment (MCI) is an early stage of cognitive decline or memory loss, commonly found among the elderly. A phonemic verbal fluency (PVF) task is a standard cognitive test that participants are asked to produce words starting with given letters, such as “F” in English and “ก” /k/ in Thai. With state-of-the-art machine learning techniques, features extracted from the PVF data have been widely used to detect MCI. The PVF features, including acoustic features, semantic features, and word grouping, have been studied in many languages but not Thai. However, applying the same PVF feature extraction methods used in English to Thai yields unpleasant results due to different language characteristics. This study performs analytical feature extraction on Thai PVF data to classify MCI patients. In particular, we propose novel approaches to extract features based on phonemic clustering (ability to cluster words by phonemes) and switching (ability to shift between clusters) for the Thai PVF data. The comparison results of the three classifiers revealed that the support vector machine performed the best with an area under the receiver operating characteristic curve (AUC) of 0.733 (N = 100). Furthermore, our implemented guidelines extracted efficient features, which support the machine learning models regarding MCI detection on Thai PVF data.
Open AccessSelecting the Most Important Features for Predicting Mild Cognitive Impairment from Thai Verbal Fluency Assessments
Mild cognitive impairment (MCI) is an early stage of cognitive decline or memory loss, commonly found among the elderly. A phonemic verbal fluency (PVF) task is a standard cognitive test that participants are asked to produce words starting with given letters, such as “F” in English and “ก” /k/ in Thai. With state-of-the-art machine learning techniques, features extracted from the PVF data have been widely used to detect MCI. The PVF features, including acoustic features, semantic features, and word grouping, have been studied in many languages but not Thai. However, applying the same PVF feature extraction methods used in English to Thai yields unpleasant results due to different language characteristics. This study performs analytical feature extraction on Thai PVF data to classify MCI patients. In particular, we propose novel approaches to extract features based on phonemic clustering (ability to cluster words by phonemes) and switching (ability to shift between clusters) for the Thai PVF data. The comparison results of the three classifiers revealed that the support vector machine performed the best with an area under the receiver operating characteristic curve (AUC) of 0.733 (N = 100). Furthermore, our implemented guidelines extracted efficient features, which support the machine learning models regarding MCI detection on Thai PVF data.
- Open Access
- 12 June 2022
- Published by Taylor & Francis in International Journal of Neuroscience
- Vol. 134 (1) , 16-27
Construction of a short version of the Montreal Cognitive Assessment (MoCA) rating scale for the Thai population using Partial Least Squares analysis
The Montreal Cognitive Assessment (MoCA) rating scale is frequently used to assess cognitive impairments in amnestic mild cognitive impairment (aMCI) and Alzheimer’s disease (AD).The aims of this s...
- 12 June 2022
- Published by Taylor & Francis in International Journal of Neuroscience
- Vol. 134 (1) , 16-27
Open AccessConstruction of a short version of the Montreal Cognitive Assessment (MoCA) rating scale for the Thai population using Partial Least Squares analysis
The Montreal Cognitive Assessment (MoCA) rating scale is frequently used to assess cognitive impairments in amnestic mild cognitive impairment (aMCI) and Alzheimer’s disease (AD).The aims of this s...
- Open Access
Using Automatic Speech Recognition to Assess Thai Speech Language Fluency in the Montreal Cognitive Assessment (MoCA)
The Montreal cognitive assessment (MoCA), a widely accepted screening tool for identifying patients with mild cognitive impairment (MCI), includes a language fluency test of verbal functioning; its scores are based on the number of unique correct words produced by the test taker. However, it is possible that unique words may be counted differently for various languages. This study focuses on Thai as a language that differs from English in terms of word combinations. We applied various automatic speech recognition (ASR) techniques to develop an assisted scoring system for the MoCA language fluency test with Thai language support. This was a challenge because Thai is a low-resource language for which domain-specific data are not publicly available, especially speech data from patients with MCIs. Furthermore, the great variety of pronunciation, intonation, tone, and accent of the patients, all of which might differ from healthy controls, bring more complexity to the model. We propose a hybrid time delay neural network hidden Markov model (TDNN-HMM) architecture for acoustic model training to create our ASR system that is robust to environmental noise and to the variation of voice quality impacted by MCI. The LOTUS Thai speech corpus was incorporated into the training set to improve the model’s generalization. A preprocessing algorithm was implemented to reduce the background noise and improve the overall data quality before feeding data into the TDNN-HMM system for automatic word detection and language fluency score calculation. The results show that the TDNN-HMM model in combination with data augmentation using lattice-free maximum mutual information (LF-MMI) objective function provides a word error rate (WER) of 30.77%. To our knowledge, this is the first study to develop an ASR with Thai language support to automate the scoring system of MoCA’s language fluency assessment.
Open AccessUsing Automatic Speech Recognition to Assess Thai Speech Language Fluency in the Montreal Cognitive Assessment (MoCA)
The Montreal cognitive assessment (MoCA), a widely accepted screening tool for identifying patients with mild cognitive impairment (MCI), includes a language fluency test of verbal functioning; its scores are based on the number of unique correct words produced by the test taker. However, it is possible that unique words may be counted differently for various languages. This study focuses on Thai as a language that differs from English in terms of word combinations. We applied various automatic speech recognition (ASR) techniques to develop an assisted scoring system for the MoCA language fluency test with Thai language support. This was a challenge because Thai is a low-resource language for which domain-specific data are not publicly available, especially speech data from patients with MCIs. Furthermore, the great variety of pronunciation, intonation, tone, and accent of the patients, all of which might differ from healthy controls, bring more complexity to the model. We propose a hybrid time delay neural network hidden Markov model (TDNN-HMM) architecture for acoustic model training to create our ASR system that is robust to environmental noise and to the variation of voice quality impacted by MCI. The LOTUS Thai speech corpus was incorporated into the training set to improve the model’s generalization. A preprocessing algorithm was implemented to reduce the background noise and improve the overall data quality before feeding data into the TDNN-HMM system for automatic word detection and language fluency score calculation. The results show that the TDNN-HMM model in combination with data augmentation using lattice-free maximum mutual information (LF-MMI) objective function provides a word error rate (WER) of 30.77%. To our knowledge, this is the first study to develop an ASR with Thai language support to automate the scoring system of MoCA’s language fluency assessment.