Sorawat Sangkaew

Dr. Sorawat Sangkaew

orcid

Research Focus & Metrics

Academic Career Length

5Years (2022-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
18 January 2022 in PLOS Digital Health
Verified
Journal Article
Open access
PDF file
DMDamien Ming
Damien Ming
BHBernard Hernandez
Bernard Hernandez
SSSorawat Sangkaew
Sorawat Sangkaew
NVNguyen Lam Vuong
Nguyen Lam Vuong
PLPhung Khanh Lam
Phung Khanh Lam
OIon behalf of the Vietnam ICU Translational Applications Laboratory (VITAL) investigators
on behalf of the Vietnam ICU Translational Applications Laboratory (VITAL) investigators
Background: Identifying patients at risk of dengue shock syndrome (DSS) is vital for effective healthcare delivery. This can be challenging in endemic settings because of high caseloads and limited resources. Machine learning models trained using clinical data could support decision-making in this context. Methods: We developed supervised machine learning prediction models using pooled data from adult and paediatric patients hospitalised with dengue. Individuals from 5 prospective clinical studies in Ho Chi Minh City, Vietnam conducted between 12th April 2001 and 30th January 2018 were included. The outcome was onset of dengue shock syndrome during hospitalisation. Data underwent random stratified splitting at 80:20 ratio with the former used only for model development. Ten-fold cross-validation was used for hyperparameter optimisation and confidence intervals derived from percentile bootstrapping. Optimised models were evaluated against the hold-out set. Findings: The final dataset included 4,131 patients (477 adults and 3,654 children). DSS was experienced by 222 (5.4%) of individuals. Predictors were age, sex, weight, day of illness at hospitalisation, indices of haematocrit and platelets over first 48 hours of admission and before the onset of DSS. An artificial neural network model (ANN) model had best performance with an area under receiver operator curve (AUROC) of 0.83 (95% confidence interval [CI], 0.76–0.85) in predicting DSS. When evaluated against the independent hold-out set this calibrated model exhibited an AUROC of 0.82, specificity of 0.84, sensitivity of 0.66, positive predictive value of 0.18 and negative predictive value of 0.98. Interpretation: The study demonstrates additional insights can be obtained from basic healthcare data, when applied through a machine learning framework. The high negative predictive value could support interventions such as early discharge or ambulatory patient management in this population. Work is underway to incorporate these findings into an electronic clinical decision support system to guide individual patient management.

9 February 2026 in PLOS Digital Health
Verified
Journal Article
Open access
PDF file
SSSorawat Sangkaew
Sorawat Sangkaew
BDBethan Cracknell Daniels
Bethan Cracknell Daniels
DMDamien K. Ming
Damien K. Ming
BHBernard Hernandez
Bernard Hernandez
PHPau Herrero
Pau Herrero
IDIlaria Dorigatti
Ilaria Dorigatti
Dengue severity prediction models are usually developed using hospitalized patient data, but triage and hospital admission are mainly evaluated in outpatient settings. This study developed models using clinical and laboratory data from patients in outpatient settings during the febrile phase. Data from two cohort studies in Vietnam and Thailand were used to develop and validate six models: logistic regression with warning signs, Lasso-selected logistic regression, random forest, extreme gradient boosted classification, support vector machine, and artificial neural network. Models predicted dengue shock syndrome (DSS) as the primary endpoint and moderate plasma leakage and/or DSS as the secondary endpoint. We assessed model performance, discrimination, and calibration, using sensitivity, specificity, accuracy, Brier score, AUROC, CITL, calibration slope, calibration plots, and decision curve analysis. The optimal model was the Lasso-selected logistic regression for predicting DSS and the combined endpoint of moderate plasma leakage and/or DSS (Brier score: 0.044 [95% CI 0.043, 0.044] and 0.104 [95% CI 0.104, 0.105]; AUROC: 0.789 [95% CI 0.787, 0.791] and 0.741 [95% CI 0.740, 0.742]). We identified hematocrit, platelet count, lymphocyte count, and aspartate aminotransferase as predictors for DSS, and abdominal pain or tenderness, vomiting, mucosal bleeding, white blood cell count, lymphocyte count, platelet count, aspartate aminotransferase, and serum albumin as predictors for the secondary endpoint. Logistic regression and machine learning models using clinical and laboratory data during the febrile phase can support early prediction of severe disease in outpatient settings. Integrating risk prediction models into a decision support system could improve triage and optimize healthcare and resource allocation in endemic and resource-limited areas.

25 July 2025 in Kidney360
Verified
Journal Article
Open access
KCKittiphan Chienwichai
Kittiphan Chienwichai
SJSirin Jiwakanon
Sirin Jiwakanon
KCKamonrat Chaiviriyawong
Kamonrat Chaiviriyawong
JWJananya Wattanakul
Jananya Wattanakul
WRWarat Rojnsaengroung
Warat Rojnsaengroung
PMPannawat Mongkolrattanakul
Pannawat Mongkolrattanakul
Key Points: The Furst equation helps identify patients at risk of fluid restriction failure in syndrome of inappropriate antidiuresis, although its predictive performance is limited. Additional urinary biomarkers beyond the Furst equation may not improve the prediction of fluid restriction response. Background: The syndrome of inappropriate antidiuresis (SIAD) is a common cause of hyponatremia in hospitalized patients, with fluid restriction (FR) as the first-line treatment. However, up to 50% of patients fail to respond to FR, highlighting the need for reliable predictors of treatment failure. Therefore, this study aimed to evaluate the predictive performance of urinary biomarkers, including the Furst equation (the ratio of the sum of urinary sodium and potassium concentrations to plasma sodium), urine osmolality, and urine output, in identifying nonresponders to FR. Methods: This prospective multicenter cohort study was conducted at two medical centers in Thailand from September 2022 to June 2024 and included 79 hospitalized patients with SIAD. Response to FR was defined as an increase in serum sodium level of ≥3 mEq/L by day 4, with no decrease in serum sodium levels at any point during the FR period. Predictive performance was assessed using sensitivity, specificity, and area under the receiver operating characteristic curve (AUC) for each biomarker. Multivariable logistic regression analysis incorporating all urinary biomarkers was also performed. Results: Overall, 59% of patients failed to respond to FR by day 4, and adherence to FR was modest (72%). The Furst equation demonstrated the best predictive performance, with an optimal cutoff of 0.86 (sensitivity: 55.3%; specificity: 70.2%, AUC: 0.65). Combining the Furst equation with urine osmolality and urine output improved specificity to 91.0%, although sensitivity decreased to 32.3%. A multivariable logistic regression model achieved similar discrimination (AUC: 0.67) compared with the Furst equation alone. Conclusions: Although the Furst equation can identify nonresponders to FR, its predictive performance is limited. Our findings suggest that other urinary biomarkers beyond the Furst equation may not improve overall predictive accuracy for predicting nonresponse to FR. These findings highlight the need for more effective predictive tools. Further studies are warranted to validate these results and optimize treatment strategies for SIAD. Clinical Trial registry name and registration number: Thai Clinical Trials Registry, TCTR20221206001.

13 September 2024 in IJID Regions
Verified
Journal Article
Open access
SMSupisara Meesoontorn
Supisara Meesoontorn
WPWisitsak Pakdee
Wisitsak Pakdee
CCChutima Cheranakhorn
Chutima Cheranakhorn
SSSorawat Sangkaew
Sorawat Sangkaew
This retrospective cohort study aims to assess the incidence and associated factors of in-hospital mortality and respiratory failure among patients with COVID-19 during the Omicron and Delta epidemics. We reviewed medical records from a regional hospital in Southern Thailand of patients with COVID-19 during the Delta wave (August 2021 to December 2021) and the Omicron wave (February 2022 to June 2022). A computer-generated random sampling approach was used to select patients for analysis. Logistic regression identified factors associated with respiratory failure, while Cox proportional hazard models were used for in-hospital mortality associations. Among 5729 hospitalized patients with COVID-19 (Delta: 1229; Omicron: 4500), 1164 were randomly sampled (Delta: 295; Omicron: 869). Patients during the Delta wave were older (median: 52.0; interquartile range: 31.5-70.0) compared with those during the Omicron wave (median: 37.0; interquartile range: 8.0-65.0), with lower rates of completed two doses of vaccination (Delta: 18.3%; Omicron: 51.8%). The Delta variant exhibited higher rates of respiratory failure (46.8% vs Omicron: 22.3%) and in-hospital mortality (12.5% vs Omicron: 6.9%). Increased age (adjusted odds ratio [aOR] 1.04, 95% confidence interval [CI] 1.03-1.04), severity at admission (aOR 12.48, 95% CI 7.21-22.22), and delayed time to admission (aOR 1.07, 95% CI 1.02-1.12) increased the risk of respiratory failure, while receiving two-dose vaccination reduced this risk (aOR 0.36, 95% CI 0.24-0.53). Similarly, increased age (adjusted hazard ratio [aHR] 1.05, 95% CI 1.03-1.06), severity at admission (aHR 7.20, 95% CI 4.09-12.60), and delayed time to admission (aHR 1.05, 95% CI 1.00-1.11) were associated with higher in-hospital mortality, with two-dose vaccination decreasing this risk (aHR 0.45, 95% CI 0.27-0.75). The Delta variant exhibited higher in-hospital mortality and respiratory failure rates compared with Omicron. The identification of high-risk groups emphasizes the critical need for timely care for vulnerable patients. Timely access to care and vaccination coverage are crucial in reducing respiratory failure and mortality due to COVID-19, highlighting the necessity for tailored interventions to mitigate the impact of emerging variants.

10 October 2024 in BMC Infectious Diseases
Verified
Journal Article
Open access
PDF file
SSSorawat Sangkaew
Sorawat Sangkaew
HTHathaitip Tumviriyakul
Hathaitip Tumviriyakul
CCChutima Cheranakhorn
Chutima Cheranakhorn
NSNopporn Songumpai
Nopporn Songumpai
NPNawamin Pinpathomrat
Nawamin Pinpathomrat
IDIlaria Dorigatti
Ilaria Dorigatti
Background: Post-COVID- 19 syndrome (PCS) significantly impacts the quality of life of survivors. There is, however, a lack of a standardized approach to PCS diagnosis and management. Our bidirectional cohort study aimed to estimate PCS incidence, identify risk factors and biomarkers, and classify clinical phenotypes for enhanced management to improve patient outcomes. Methods: A bidirectional prospective cohort study was conducted at five medical sites in Hatyai district in Songkhla Province, Thailand. Participants were randomly selected from among the survivors of COVID-19 aged≥18 years between May 15, 2022, and January 31, 2023. The selected participants underwent a scheduled outpatient visit for symptom and health assessments 12 to 16 weeks after the acute onset of infection, during which PCS was diagnosed and blood samples were collected for hematological, inflammatory, and serological tests. PCS was defined according to the World Health Organization criteria. Univariate and multiple logistic regression analyses were used to identify biomarkers associated with PCS. Moreover, three clustering methods (agglomerative hierarchical, divisive hierarchical, and K-means clustering) were applied, and internal validation metrics were used to determine clustering and similarities in phenotypes. Findings: A total of 300 survivors were enrolled in the study, 47% of whom developed PCS according to the World Health Organization (WHO) definition. In the sampled cohort, 66.3% were females, and 79.4% of them developed PCS (as compared to 54.7% of males, p-value <0.001). Comorbidities were present in 19% (57/300) of all patients, with 11% (18/159) in the group without PCS and 27.7% (39/141) in the group with PCS. The incidence of PCS varied depending on the criteria used and reached 13% when a quality of life indicator was added to the WHO definition. Common PCS symptoms were hair loss (22%) and fatigue (21%), while mental health symptoms were less frequent (insomnia 3%, depression 3%, anxiety 2%). According to our univariate analysis, we found significantly lower hematocrit and IgG levels and greater ALP levels in PCS patients than in patients who did not develop PCS (p-value < 0.05). According to our multivariable analysis, adjusted ALP levels remained a significant predictor of PCS (OR 1.02, p-value= 0.005). Clustering analysis revealed four groups characterized by severe clinical symptoms and mental health concerns (Cluster 1, 4%), moderate physical symptoms with predominant mental health issues (Cluster 2, 9%), moderate mental health issues with predominant physical symptoms (Cluster 3, 14%), and mild to no PCS (Cluster 4, 77%). The quality of life and ALP levels varied across the clusters. Interpretation: This study challenges the prevailing diagnostic criteria for PCS, emphasizing the need for a holistic approach that considers quality of life. The identification of ALP as a biomarker associated with PCS suggests that its monitoring could be used for early detection of the onset of PCS. Cluster analysis revealed four distinct clinical phenotypes characterized by different clinical symptoms and mental health concerns that 'exhibited varying impacts on quality of life. This finding suggested that accounting for the reduced quality of life in the definition of PCS could enhance its diagnosis and management and that moving toward personalized interventions could both improve patient outcomes and help reduce medicalization and optimally target the available resources. Funding: The research publication received funding support from Medical Council of Thailand (Police General Dr. Jongjate Aojanepong Foundation), Hatyai Hospital Charity and Wellcome Trust.

1 January 2025 in The ASEAN Journal of Radiology
Verified
Journal Article
Open access
PDF file
WPWisitsak Pakdee
Wisitsak Pakdee
SSSorawat Sangkaew
Sorawat Sangkaew
RWRichard Wilson
Richard Wilson
PTPramot Tanutit
Pramot Tanutit
Over recent decades, machine learning has been widely implemented in medical radiology. Radiologists, who are at the forefront of clinical practice, need to be aware of the benefits of machine learning to facilitate its implementation. It is crucial for them to thoroughly understand and effectively integrate machine learning into the practical realm of medical radiology. In this review, we highlight the principles and applications of machine learning in medical radiology and provide a summary of its development in this field. Machine learning has significantly advanced diagnostic imaging, enhancing detection, segmentation, and image reconstruction, while improving workflow efficiency and radiology reporting. Current literature indicates three primary challenges in implementing machine learning: data standardization, validation of model performance, and regulatory compliance. The successful integration of machine learning in clinical practice requires robust data security protocols and clear frameworks for professional accountability. To prepare for this technological transition, radiologists must develop new competencies through enhanced educational programs and adapt their roles to focus more on clinical decision-making and multidisciplinary collaboration while leveraging machine learning as a supportive tool.