NPNatthaphon Pruksathorn

Natthaphon Pruksathorn

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Natthaphon Pruksathorn has a total of 22 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Comparison of three emergency department triage systems among Thai medical students: A randomized three-period crossover simulation study

    by Jarupa Yaowalaorng, Thanin Lokeskrawee, Natthaphon Pruksathorn, Suppachai Lawanaskol, Jayanton Patumanond, Chanitda Wicha, Wanwisa Bumrungpagdee, Suwapim Chanlaor, Chawalit Lakdee

    Introduction: In Thailand, the Emergency Severity Index (ESI) and Ministry of Public Health Emergency Department triage system (MOPH) are commonly used, whereas the Thai Rapid Initial Assessment and Grading for Emergency (TRIAGE) system has been newly proposed. Comparative evidence is limited. We compared correct triage as the primary outcome, with under-triage, over-triage, and completion time as secondary outcomes. Methods: We conducted a randomized three-period crossover simulation study at the Lampang Medical Education Center from October 2025 to January 2026. Fifth- and sixth-year medical students without recent standardized triage reinforcement classified 50 standardized patient cards per system in one of six sequences, with a one-week washout between periods. Multilevel mixed-effects models were used to estimate between-system differences, with adjustment for period, sequence, and academic year. Results: Sixty students completed all periods. No significant overall differences were detected in correct triage across TRIAGE, ESI, and MOPH (63.2%, 64.5%, and 61.8%, respectively), or in under-triage (19.6%, 20.0%, and 18.8%). Among sixth-year students, correct triage differed significantly across systems (global P = 0.004), with higher adjusted probabilities for ESI and TRIAGE than MOPH. Over-triage differed across systems and was highest with MOPH (7.0%), followed by TRIAGE (5.7%) and ESI (4.9%). Completion time was longest with TRIAGE (32.8 minutes), compared with ESI (24.3 minutes) and MOPH (23.2 minutes). Conclusion: No significant overall differences in correct or under-triage were detected, but the study was not powered to establish equivalence. MOPH showed higher over-triage, Completion time was longest with TRIAGE. Generalizability to experienced clinicians requires further study. Trial registration: Thai Clinical Trials Registry (TCTR) TCTR20260316003

  • Open Access

    Predicting Breast Imaging-Reporting and Data System Classification of Palpable Breast Masses Using Ultrasound to Prioritize Mammography Queues

    by Sarisa Thinyu, Thanin Lokeskrawee, Takumi Sakata, Natthaphon Pruksathorn, Suppachai Lawanaskol, Jayanton Patumanond, Suwapim Chanlaor, Wanwisa Bumrungpagdee, Chawalit Lakdee

    Background: Breast cancer is the leading cause of cancer death in women worldwide. Breast imaging, usually mammography and/or ultrasound, is classified using the Breast Imaging-Reporting and Data System (BI-RADS). At Lampang Hospital, mammography delays of up to 5 months postpone diagnosis in 40% of breast cancer cases. An urgent queue for palpable breast masses was introduced, but nearly half were benign, leading to inefficient prioritization. This study aimed to develop a two-step model based on high-risk ultrasound features and compare it with reference BI-RADS classifications. Methods: This diagnostic prediction study collected retrospective data from Lampang Hospital between January 2021 and December 2023. Ultrasound images of 390 patients were independently reviewed by radiologists blinded to the reference BI-RADS classification. Stepwise multivariable risk difference regression analysis was applied to identify predictive characteristics from seven predefined ultrasound findings. Results: Three predictive characteristics were identified: shape, margin, and echo pattern. The two-step model showed excellent discrimination, with an area under the receiver operating characteristic curve (AuROC) of 0.9801 (95% CI, 0.9696–0.9907) in step 1 and 0.9623 (95% CI, 0.9411–0.9835) in step 2. Internal validation with 200 bootstrap cycles confirmed minimal optimism. Using prevalence-based cut points, the model achieved 88.5% accuracy, with 6.7% underestimation in BI-RADS 4–5 (predicted as 3) and overestimation not exceeding 3% in any category. Conclusions: A two-step ultrasound-based model using shape, margin, and echo pattern demonstrated excellent discrimination as well as high accuracy, with slightly increased underestimation and minimal overestimation. This re-scheduling strategy optimizes mammography queue prioritization, but external validation is required before clinical implementation.

  • Open Access

    Mortality Following Intracerebral Hemorrhage After Intravenous Alteplase in Acute Ischemic Stroke: A Decision Tree-Based Classification Approach

    by Bongkot Wijitjarassang, Thanin Lokeskrawee, Natthaphon Pruksathorn, Jarupa Yaowalaorng, Suppachai Lawanaskol, Jayanton Patumanond, Wanwisa Bumrungpagdee, Suwapim Chanlaor, Chawalit Lakdee

  • Open Access

    Machine Learning-Based Model to Classify Emergency Severity Index Levels 1-3 in Febrile Patients With Tachycardia: Thailand Triage Prediction System

    by Chanitda Wicha, Thanin Lokeskrawee, Sagoontee Inkate, Natthaphon Pruksathorn, Jarupa Yaowalaorng, Suppachai Lawanaskol, Jayanton Patumanond, Suwapim Chanlaor, Wanwisa Bumrungpagdee, Chawalit Lakdee

    Background: Febrile patients with tachycardia present diverse profiles that complicate triage. Although the Emergency Severity Index (ESI) is widely used in Thailand, inter-rater variability limits consistency. Machine learning (ML) may enhance reliability using routinely collected triage data. The objectives were to develop and evaluate ML models predicting ESI levels 1-3 in febrile tachycardic adults and identify the best model for clinical use. Methods: This diagnostic prediction study analyzed adults with fever (≥ 37.6 °C) and pulse rate > 100 beats per minute in the triage area of Lampang Hospital, Thailand, during June - August 2024. Patients with complete data were included, whereas referrals and expert-disagreement cases were excluded. Expert-assigned ESI levels were the outcome. Thirteen routinely collected triage variables were evaluated as candidate predictors. The dataset (n = 500) was randomly split 80:20 into development and testing sets. Random forest, extreme gradient boosting (XGBoost), and gradient boosting machine models were developed using five-fold cross-validation with class-weighting for imbalance correction. Performance was assessed using area under the receiver operating characteristic curve (AuROC), calibration, and confusion matrices, with attention to clinically relevant misclassification. Results: XGBoost demonstrated the best discrimination with AuROC values of 1.00 (confidence interval (CI): 0.99 - 1.00), 0.94 (CI: 0.89 - 0.98), and 0.97 (CI: 0.93 - 1.00) for ESI levels 1-3 in the test set. Calibration showed the lowest Brier scores, and misclassification was minimal, supporting strong predictive consistency across categories. Conclusions: XGBoost was selected for integration into the Smart ER system as the Thailand Triage Prediction System (TTPS), providing real-time prediction to enhance triage accuracy, support decision-making, and improve workflow.


  • Open Access

    Comparison of the SeizCT Primer and Optimized Models for Predicting Positive Computed Tomography Findings in Patients With Non-Traumatic Seizures

    by Natthaphon Pruksathorn, Thanin Lokeskrawee, Jarupa Yaowalaorng, Suppachai Lawanaskol, Jayanton Patumanond, Suwapim Chanlaor, Wanwisa Bumrungpagdee, Chawalit Lakdee, Kreshya Sudsanoh, Pimploy Suriyanusorn

    Background: Previous studies developed the SeizCT primer and optimized models, both demonstrating similar values for the area under the receiver operating characteristic curve (AuROC). The optimized model incorporates Glasgow Coma Scale (GCS) change from baseline instead of categorized GCS at emergency department (ED) presentation. This study aimed to validate these two models for predicting positive computed tomography (CT) findings in patients with non-traumatic seizures. Methods: A retrospective cross-sectional study was conducted among adult patients (≥ 18 years) with non-traumatic seizures who underwent CT brain imaging in the ED at Lampang Hospital. Data were collected between December 2023 and July 2024 based on parameters from the SeizCT primer and optimized models. External validation compared model performance using AuROC, calibration, decision curve analysis (DCA), and confusion matrices. Results: The validation cohort included 312 patients (210 (67.3%) male; mean age 53 years). Positive CT findings were found in 58 patients (18.6%). The SeizCT primer model had an AuROC of 0.7218 (95% confidence interval (CI): 0.6476, 0.7961), while the SeizCT optimized model achieved 0.7394 (95% CI: 0.6663, 0.8124). An equivalence test showed statistically equivalent discrimination between the two models (P < 0.015), with the optimized model demonstrating better calibration (slope: 0.582 vs. 0.701; P = 0.002; observed-to-expected ratio: 0.934 vs. 0.971; P = 0.030). Conclusions: Both models demonstrated fair to acceptable discrimination after external validation. The SeizCT optimized model is recommended, as it showed superior calibration and its incorporation of GCS change from baseline offers greater clinical applicability.

  • Open Access

    Effectiveness of Short-Acting Oral Antihypertensives in Asymptomatic Markedly Elevated Blood Pressure: An Inverse Probability Treatment Weighted Study

    by Supark Chotichaipiboon, Thanin Lokeskrawee, Natthaphon Pruksathorn, Jarupa Yaowalaorng, Suppachai Lawanaskol, Jayanton Patumanond, Wanwisa Bumrungpagdee, Suwapim Chanlaor, Chawalit Lakdee

    Background: Although guidelines recommend gradual blood pressure (BP) reduction in asymptomatic markedly elevated BP (AMEBP), short-acting oral antihypertensives are often used despite limited evidence - driven by healthcare providers’ concerns or institutional policies - thereby increasing system burden. This study aimed to identify practical regimens enabling safe, timely discharge and continuity of care. Methods: This retrospective cohort included patients with AMEBP (systolic BP > 180 mm Hg and/or diastolic BP > 110 mm Hg), who presented to the emergency department (ED) of Lampang Hospital between January and December 2023. Oral regimens included captopril, hydralazine, metoprolol (25 mg each), or combinations. The outcome was mean arterial pressure (MAP) reduction within 60 min, classified as no reduction (< 10%), appropriate reduction (10-15%), or over-reduction (> 15%). Inverse probability of treatment weighting (IPTW) adjusted for regimen selection, and multinomial odds ratios (mORs) with 95% confidence intervals (CIs) were reported. Results: Among 158 patients (mean age 60 years; 67.1% female), baseline MAP was 142.4 ± 16.2 mm Hg. Captopril and hydralazine monotherapy were most likely to achieve appropriate reduction (captopril: mOR: 2.91; 95% CI: 0.96, 8.76; P = 0.058; hydralazine: mOR: 2.76; 95% CI: 0.84, 9.11; P = 0.096). Metoprolol was associated with inappropriate response, while captopril plus hydralazine had highest odds of over-reduction (mOR: 3.04; 95% CI: 1.02, 9.00; P = 0.045). Conclusions: Although urgent BP reduction is not routinely recommended in the ED, oral captopril appears to be a reasonable first-line option when clinically necessary, while hydralazine may be a suitable alternative. Combination of these two drugs should be used with caution due to the risk of excessive MAP reduction.

  • Open Access

    Optimal Timing of Post-Alteplase Brain Computed Tomography: Routine Twenty-Four-Hour Versus Extended Forty-Eight-Hour for Detecting Asymptomatic Intracranial Hemorrhage

    by Pranisa Botkosa, Thanin Lokeskrawee, Donsuk Pongnikorn, Natthaphon Pruksathorn, Suppachai Lawanaskol, Jayanton Patumanond, Jarupa Yaowalaorng, Suwapim Chanlaor, Wanwisa Bumrungpagdee, Chawalit Lakdee

    Background: Intracranial hemorrhage (ICH) is a major complication of intravenous alteplase for acute ischemic stroke. Asymptomatic ICH (asICH) does not cause immediate neurological decline but may have medicolegal consequences if undetected. In Asian populations, up to 24% of post-alteplase ICH occurs beyond 24 h, suggesting that routine computed tomography (CT) at 24 h may underestimate its incidence. This study evaluated the diagnostic yield of CT at 24 h versus 48 h to determine the optimal timing for ICH detection. Methods: A retrospective cohort study was conducted at Lampang Hospital from March 2017 to December 2024. Patient outcomes were analyzed using an illness-death multistate model with three health states: no ICH, asICH, and symptomatic ICH (sICH). Three transitions were modeled - from no ICH to asICH, from no ICH to sICH, and from asICH to sICH - assuming irreversible progression. The proportions of asICH detected by CT at 24 h versus 48 h were compared using the exact probability test. Results: After exclusions, a total of 555 patients were eligible (54.8% male, mean age 65 years). Final transition outcomes identified 455 patients with no ICH, 51 with asICH, and 49 with sICH. Brain CT at 24 h detected 30.2% of asICH cases, while 48-h imaging detected 96.8%, with a significant improvement (P < 0.001). Conclusions: CT imaging at 48 h post-alteplase improves asICH detection compared with the 24-h scan, reducing missed cases without added radiation or resource burden. Multicenter validation is needed to confirm whether 48 h should replace the current standard.

  • Open Access

    Effectiveness of Antibiotic Regimens in Reducing White Blood Cell Count Within Three to Five Days in Febrile Leukocytosis Treated With Ambulatory Therapy

    by Woraphat Tumporn, Thanin Lokeskrawee, Natthaphon Pruksathorn, Suppachai Lawanaskol, Jayanton Patumanond, Suwapim Chanlaor, Wanwisa Bumrungpagdee, Chawalit Lakdee

    Background: In the absence of sepsis, patients with fever and leukocytosis in the emergency department (ED) are often treated with ambulatory parenteral antibiotics at the physician’s discretion. Identifying effective regimens for reducing white blood cell (WBC) count and improving clinical outcomes may support standardized ED care. Methods: This retrospective cohort included adult ED patients with fever and leukocytosis, but without sepsis, in whom basic investigation revealed no clear source of infection. Patients received one of three regimens: (A) single-day intravenous (IV) ceftriaxone followed by oral cefixime, (B) 3-day IV ceftriaxone followed by oral cefixime, or (C) 3-day IV ceftriaxone plus oral doxycycline from day 1. Demographics, baseline data, and laboratory results were collected. Follow-up assessments included WBC count and clinical improvement. Treatment probabilities were estimated with multinomial logistic regression, and stabilized inverse probability of treatment weighting (IPTW) were applied in weighted quantile regression. Results: Among 250 patients, most were female (63.2%) with a mean age of 49 years. After adjustment, regimen C showed a median WBC reduction of -6.9 × 103 cells/mm3 (95% confidence interval (CI): -7.5 to -6.3), compared with -6.8 × 103 (95% CI: -7.2 to -6.5) for regimen B and -6.6 × 103 (95% CI: -7.0 to -6.2) for regimen A. Differences were not significant (Wald test, P = 0.484), but graphical analysis suggested the steepest decline with regimen C. Conclusion: Though regimen C showed the steepest WBC decline and fewer failures, the study was markedly underpowered (< 10%). Larger multicenter studies are required to confirm these findings.

  • Open Access

    Improving a Clinical Prediction Model for Computed Tomography Head Scan Use in Non-Traumatic Seizures: The SeizCT Optimized Model

    by Kreshya Sudsanoh, Thanin Lokeskrawee, Natthaphon Pruksathorn, Suppachai Lawanaskol, Jayanton Patumanond, Wanwisa Bumrungpagdee, Suwapim Chanlaor, Chawalit Lakdee, Pimploy Suriyanusorn

    Background: A previous study developed a predictive model for identifying positive findings on computed tomography (CT) brain imaging in patients with non-traumatic seizures. One key predictor was the Glasgow Coma Scale (GCS), categorized into three groups (≤ 8, 9 - 13, >13). However, in real-world practice, some bedridden patients have low baseline GCS, which may lead to overprediction and unnecessary CT imaging. The original model yielded a false positive of 16.6%, exceeding the predefined threshold of less than 15%. This study aimed to improve the model by replacing categorized GCS with GCS change from baseline to reduce false positives below 15%, while maintaining false negatives below 5%. Methods: This diagnostic predictive study included patients with non-traumatic seizures who underwent CT brain imaging at the emergency department between November 2019 and November 2023. The original predictors were retained, with GCS change from baseline replacing the categorized GCS. Multivariable logistic regression analysis was used to estimate multivariable odds ratios. Discriminative performance was assessed using the area under the receiver operating characteristic (AuROC) curve. A trade-off between sensitivity and specificity was applied to identify a probability cut-off that met the target for false positives and false negatives. The revised model was named the “SeizCT optimized model”. Results: Of the 625 patients included, 18.9% had positive CT findings. The majority were male (74.9%) with a mean age of 55 years. The SeizCT optimized model incorporated six predictors: prior stroke (> 3 months), current cancer, GCS change from baseline, alcohol withdrawal symptoms, epilepsy, and focal neurological deficit. The model demonstrated an AuROC of 0.8221 (95% confidence interval (CI): 0.7813, 0.8629). Using a threshold probability of 22.57%, it outperformed the original model (AuROC of 0.8156; 95% CI: 0.7598, 0.8746) with narrower confidence intervals. It achieved a false negative of 4.8% and a false positive of 14.4%. Conclusions: The SeizCT optimized model showed improved performance over the original tool, reducing both false positives and false negatives within the predefined thresholds. External validation is recommended.