Chulalongkorn Medical Journal

Supervised learning algorithms for subject- and textlevel classification of depression-related context among Thai-language Facebook users

2026-07-21T13:08:50+07:00

Background: Depression is an increasing concern worldwide that requires urgent attention, particularly in Thailand. Without timely intervention, it can lead to serious consequences, such as health complications, reduced social participation, and even suicide. Consequently, research has increasingly focused on developing algorithms to identify and classify social media data into depression severity.

Objectives: This study aimed to examine label mismatches between self-assessment forms and social media behaviors and to evaluate the feasibility of using experts’ opinions to classify Thai Facebook text into depression categories.

Methods: The methodology involved data collection through surveys and Facebook scraping, followed by data preprocessing, feature engineering, and model training and validation. The models employed included the support vector machine, logistic regression, and decision tree, applied at the subject and text levels. Labels were derived from the Thai Depression Inventory (TDI) and experts’ evaluations.

Results: Only 30 out of 60 participants actively used Facebook during the study period, with substantial variation in posting behavior. Subject-level classification using TDI labels yielded better, more balanced performance metrics than classifications based on experts’ opinions at the subject and text levels. Part-ofspeech tagging showed slight correlations with depression severity, whereas emotions did not.

Conclusion: Subject-level classification using self-assessment labels performed better than expert-derived labels and text-level classification when compared to a dummy baseline model. This suggests a slight correlation between the self-assessment forms and social media behaviors. In contrast, the poor agreement between labeling approaches suggests that text data alone may be insufficient as a reliable clinical indicator of depression.

Machine learning-based gene signature detection highlights CXCL12 as a key marker for acute myeloid leukemia prediction

2026-07-14T08:58:03+07:00

Background: Acute myeloid leukemia (AML) is a severe hematologic malignancy that is marked by the uncontrolled proliferation and impaired differentiation of myeloid cells, which disrupts normal hematopoiesis and results in poor clinical outcomes. Conventional diagnostic approaches often lack the precision needed to accurately classify AML subtypes; therefore, integrating advanced computational methods is required to improve diagnostic and therapeutic strategies.

Objectives: This study aimed to apply machine learning (ML) techniques to transcriptomic data to identify a concise, informative gene signature that can distinguish AML cases from normal samples. In addition, the study sought to evaluate the performance of predictive models in supporting AML prediction and facilitating biomarker discovery.

Methods: Gene expression data were obtained from the TARGET-AML project via the Genomic Data Commons portal. Feature selection was performed using SelectKBest with chi-square scoring to identify the top 10 AML-associated genes. To address class imbalance, SMOTE was employed. Several ML models, including Random Forest, XGBoost, LightGBM, and a Stacking ensemble, were trained and optimized via hyperparameter tuning to classify AML vs. normal samples.

Results: The selected gene panel included CXCL12, SELENBP1, SLC4A1, IFIT1B, ALAS2, CAMP, OLFM4, DEFA3, CRISP3, and HBG1, which are functionally associated with immune response, inflammation, and hematopoietic processes. All models demonstrated robust classification performance: Random Forest (88.6%), XGBoost (88.8%), LightGBM (89.4%), and Stacking (89.4%). SMOTE effectively enhanced the model performance, particularly for underrepresented classes, thereby improving precision and recall.

Conclusion: This study identified a small, yet informative gene set that accurately differentiates AML from normal samples, with LightGBM revealing the highest predictive accuracy. The involvement of immune and inflammatory genes aligns with AML’s biological foundations. These findings suggest the clinical potential of ML-based tools for AML diagnostics. Future validation on independent cohorts and the integration of multi-omic data could further improve the model’s robustness and facilitate broader applications in precision medicine for hematologic malignancies.

Ameliorative effect of the ethanolic extract of Nardostachys jatamansi against high-fat diet-induced dyslipidemia in Wistar albino rats

2026-07-09T14:26:08+07:00

Background: Dyslipidemia is a key contributor to metabolic disorders such as oxidative stress and is a major risk factor for cardiovascular diseases. Herbal interventions are increasingly preferred due to fewer side effects. Nardostachys jatamansi has demonstrated antioxidant and lipid-lowering properties, but its effects on high-fat diet (HFD)-induced dyslipidemia remain insufficiently studied.

Objective: The present study aimed to evaluate the antioxidant and hypolipidemic effects of the ethanol extract of N. jatamansi in an HFD-induced obesity model in male Wistar albino rats.

Methods: Thirty-six male Wistar albino rats were randomly divided into 6 groups, each comprising 6 animals. Group I received a normal diet for 9 weeks, whereas Group II was fed an HFD for the same period. Groups III– VI were also fed an HFD throughout the investigation period. From week 5, Group III received rosuvastatin (10 mg/kg), and Groups IV–VI were treated with N. jatamansi extract at 100, 200, and 300 mg/kg, respectively, for 4 weeks. All treatments were administered orally. At the end of the study, the animals were fasted overnight, anesthetized, and blood was collected for lipid and hematological analyses. In addition, the organs were excised for histological evaluation.

Results: N. jatamansi (100, 200, and 300 mg/kg) reversed HFD-induced increases in body weight, fat pad size, organ weight, total cholesterol, triglycerides, low-density lipoprotein, and very low-density lipoprotein levels while restoring high-density lipoprotein levels in a dose-dependent manner. Moreover, elevated white blood cells, lymphocyte counts, and serum urea, creatinine, and uric acid levels were normalized. Histological improvements in the liver suggest antioxidant and bioactive effects of N. jatamansi.

Conclusion: Our results revealed that the herbal drug N. jatamansi effectively ameliorates HFD-induced dyslipidemia through its potent antioxidant activity, thereby reducing oxidative stress and restoring lipid homeostasis.

A mathematical model for predicting blood volume to be processed for optimizing peripheral blood CD34+ cell yield during stem cell apheresis

2026-08-17T15:51:13+07:00

Background: Accurately predicting stem cell yields is important for planning peripheral blood stem cell (PBSC) collection.

Objective: This study aimed to develop a formula to forecast the blood volume required to process CD34+ cells to ensure adequate products.

Methods: Data from donors and patients who underwent PBSC apheresis at a tertiary care hospital were reviewed. BD FACSCanto™ II and Navios EX Flow Cytometry machines were used to determine the precollection CD34+ cell counts (pre-CD34).

Results: In 50 donors and 85 patients, 145 leukapheresis procedures were performed. Linear regression models accurately predicted the absolute CD34+ cell yields from pre-apheresis values, supporting the optimal planning of the procedure. Subgroup analysis identified gender, but not body weight, as a significant factor influencing collection efficiency, with lower yields in females. To ensure adequate stem cell yield, genderspecific formulas were derived to calculate the required blood volume processing. For males: Volume process (L) = (Dose CD34 [106cells/kg] * body weight recipient [kg] + 21)/ 0.44 * PreCD34 (cells/μL); For females: Volume process (L) = (Dose CD34 [106Cells/kg] * body weight recipient [kg]) – 48)/0.28 * PreCD34 (cells/μL). Furthermore, a strong correlation (r = 0.981) was observed between the two enumeration platforms, with Bland-Altman analysis confirming minimal bias and narrow variability. No significant variation in preCD34 was detected.

Conclusion: Applying preCD34 count and gender-specific predictive modeling offers valuable tools to enhance procedural efficiency and generate appropriate processed volume for the collection of adequate quantities of stem cells. Furthermore, the significance of standardizing CD34+ enumeration methods was underscored.

Integrated analysis of bioinformatic databases to identify thyroid cancer drug target candidates

2026-08-14T15:02:16+07:00

Background: Thyroid cancer (TC) is an endocrine malignancy whose global incidence continues to rise. Although it generally has a favourable prognosis, aggressive subtypes, such as anaplastic thyroid cancer, remain a therapeutic challenge. Drug repurposing strategies offer the opportunity to accelerate the development of therapies by utilising available drugs with an established safety profile.

Objectives: This study aimed to identify candidate thyroid cancer risk-associated genes by analyzing GWAS and PheWAS data and to evaluate their potential therapeutic relevance through drug recycling.

Methods: Significant gene variants were obtained from these databases and functionally annotated using HaploReg, ClinVar, SIFT, PolyPhen, CADD, cis-eQTL, and a review of PubMed literature. Candidate genes from seven functional annotations with scores ≥ 2 were designated as risk-associated and further analyzed using the STRING database. Drug candidates were searched for in DrugBank and ClinicalTrials.gov.

Results: Ten risk-associated genes may be involved in thyroid cancer pathogenesis, with DIRC3 having the highest score. However, no drugs that could directly interact with these genes were recognized. These findings highlight the therapeutic gap between thyroid cancer risk-associated genes and available drugs.

Conclusion: This study identified ten risk genes involved in thyroid cancer pathogenesis, with DIRC3 as the topmost candidate. Although no approved drugs directly targeted these genes, the findings highlight their potential as biomarkers and novel therapeutic targets. Integrative genomics and drug repurposing approaches remain promising strategies for advancing precision therapy in thyroid cancer.

Factors associated with colorectal cancer screening decision-making among older adults at the Comprehensive Geriatric Clinic, King Chulalongkorn Memorial Hospital, Thai Red Cross Society

2026-08-13T15:00:23+07:00

Background: The incidence of colorectal cancer (CRC) in Thailand is constantly increasing and is highest in individuals aged 60-75 years. However, the number of older individuals undergoing CRC screening remains suboptimal.

Objectives: This study aimed to identify factors associated with CRC screening decision-making among Thai older adults attending the Comprehensive Geriatric Clinic at King Chulalongkorn Memorial Hospital. The secondary objective was to identify factors associated with the choice of screening method—colonoscopy versus fecal occult blood testing (FOBT).

Methods: A cross-sectional study was conducted using a data record form at the Comprehensive Geriatrics Clinic, King Chulalongkorn Memorial Hospital in Bangkok, Thailand, from September to October 2023. Participants aged 60-75 were interviewed to collect information about factors associated with CRC screening decision-making, which were categorized into five groups: demographic, health, psychological, physician, and social factors. Participants were classified into two groups based on their intention to undergo CRC screening. The factors preliminarily associations using the Chi-square test and the Mann-Whitney U test. The factors with a P < 0.2 were then analysed by a binary logistic regression for identifying factors that were significantly associated with the decision to undergo CRC screening as primary outcome and the choice of method as secondary outcome.

Results: A total of 424 valid responses were obtained, with 320 respondents intending to undergo CRC screening. The binary logistic regression analysis revealed significant associations between CRC screening decision and the following factors: high risk for CRC (aOR 6.05, 95%CI 2.42 15.13), positive attitude towards cancer screening (aOR 3.65, 95%CI 2.19 6.10), recognition of the importance of early detection (aOR 2.33, 95%CI 1.32 4.11), and being divorced, as compared to being married (aOR 0.06, 95%CI 0.01 0.49). The decision to undergo colonoscopy for the CRC screening was associated with an education level of having at least a bachelor’s degree (aOR 0.43, 95%CI 0.24 0.75), being single, as compared to being married (aOR 2.08, 95%CI 1.20 3.58), and concerns about physical discomfort from CRC screening (aOR 0.41, 95%CI 0.19 0.90).

Conclusion: Factors associated with CRC screening decision-making uptake amongst older adults, attending a geriatric clinic in a tertiary care hospital in Bangkok, that can be promoted include having a positive attitude towards cancer screening and the recognition of the importance of early detection. Therefore, raising awareness and highlighting the importance of cancer screening may help increase the rates of CRC screening in Thailand in the future.

Evaluation of MDM2 mRNA expression according to clinicopathological characteristics in patients with oral squamous cell carcinoma

2026-08-18T13:24:50+07:00

Background: The murine double minute 2 (MDM2) gene encodes an E3 ubiquitin ligase that targets p53 for degradation, which influences the pathogenesis of oral squamous cell carcinoma (OSCC).

Objective. This study determined MDM2 mRNA expression and evaluated its association with the clinical and pathological characteristics thereof in patients with OSCC.

Methods: We performed an observational–analytical study using a cross-sectional design involving 39 patients with OSCC. RNA extraction, complementary DNA synthesis, and quantitative real-time polymerase chain reaction were performed. The relative MDM2 mRNA expression was quantified using the Livak method (2-ΔΔCT) with GenEX MultiD software version 6.0. Statistical significance was assessed using unpaired Student t-tests.

Results: MDM2 mRNA expression was 1.64-fold lower in patients with OSCC and tumor volumes >5 cc compared with those ≤ 5 cc (CT mean ± SE: 32.0 ± 2.6vs. 32.7 ± 1.8, respectively; P = 0.34). In high-grade cases, the expression was 0.92-fold lower than in low-grade cases (32.1 ± 2.3 vs. 32.2 ± 2.7, respectively; P = 0.90). The expression thereof was also 2.12-fold higher in smokers than in non-smokers (31.8 ± 2.4 vs. 32.9 ± 2.1, respectively; P = 0.3). None of these differences were significant. Network analysis identified MDM2, TP53, and CDKN1A as central hub genes, thus bridging p53 signaling, cell cycle regulation, and cellular senescence in OSCC. Gene Ontology enrichment revealed significant overrepresentation of p53-mediated DNA damage response and apoptotic pathways among the candidate genes.

Conclusion: Although MDM2 mRNA expression exhibited no significant clinical associations, in silico profiling identified MDM2 as a central molecular hub linking p53 signaling and cell cycle regulation in OSCC, thereby suggesting a mechanistic role that warrants further functional validation.

Beyond lipid lowering: Pleiotropic effects of statins in inflammation, coagulation, and cell proliferationmechanisms and clinical implications

2026-07-31T09:43:06+07:00

Statins are widely used lipid-lowering agents that substantially reduce cardiovascular morbidity and mortality by inhibiting 3-hydroxy-3-methylglutaryl coenzyme A reductase. Beyond their cholesterol-lowering properties, increasing evidence indicates that statins exert several pleiotropic effects that are independent of lipid reduction. These effects include anti-inflammatory, antithrombotic, and antiproliferative actions mediated primarily via the inhibition of the mevalonate pathway and downstream signaling molecules such as Rho, Rac, and Ras. Experimental and clinical studies have suggested potential benefits of statins in various noncardiovascular conditions, including autoimmune diseases, venous thromboembolism, sepsis-associated coagulopathy, and certain malignancies. In addition, pharmacoepidemiologic studies have provided insights into the real-world effects and safety profile of statins, including the risk of new-onset diabetes mellitus. This review summarizes the current evidence on the molecular mechanisms and clinical implications of statin pleiotropic effects beyond its lipid-lowering properties. However, much of the evidence supporting noncardiovascular benefits is derived from observational studies. Therefore, further randomized controlled trials are required to clarify causality, optimal patient selection, and therapeutic strategies for statin repurposing.

Identifying Pyle’s metaphyseal dysplasia: An incidental finding of skeletal pathology on a dental checkup

2026-07-13T12:48:46+07:00

Pyle’s disease, also known as familial metaphyseal dysplasia, is a rare autosomal recessive genetic disorder that primarily affects the skeletal system, particularly in children and adolescents. Furthermore, it is characterized by distinctive radiographic features, including Erlenmeyer flask deformity and metaphyseal widening. Pyle’s disease often goes underdiagnosed because of its mild clinical manifestations. This case report details a 15- year-old female patient presenting with retained milk teeth and unerupted permanent teeth, alongside notable skeletal abnormalities, including genu valgum and kyphosis. Clinical and radiographic evaluations revealed considerable dental anomalies, including multiple impacted permanent teeth and mild dysmorphic facial features. Although genetic testing was not performed in this case, a diagnosis of Pyle’s disease was made based on the clinical presentation and radiographic findings. This case highlights the importance of recognizing dental manifestations as potential indicators of underlying skeletal dysplasias or other systemic conditions. Prompt diagnosis and early intervention could enhance the quality of life and treatment outcomes in such cases.