Automated Artificial Intelligence-Based Thai Food Dietary Assessment System: Development and Validation.

Background: Dietary assessment is a fundamental component of nutrition research and plays a pivotal role in managing chronic diseases. Traditional dietary assessment methods, particularly in the context of Thai cuisine, often require extensive training and may lead to estimation errors. Objectives: To address these challenges, Institute of Nutrition, Mahidol University (INMU) iFood, an innovative artificial intelligence-based Thai food dietary assessment system, allows for estimating the nutritive values of dishes from food images. Methods: INMU iFood leverages state-of-the-art technology and integrates a validated automated Thai food analysis system. Users can use 3 distinct input methods: food image recognition, manual input, and a convenient barcode scanner. This versatility simplifies the tracking of dietary intake while maximizing data quality at the individual level. The core improvement in INMU iFood can be attributed to 2 key factors, namely, the replacement of Yolov4-tiny with Yolov7 and the expansion of noncarbohydrate source foods in the training image data set. Results: This combination significantly enhances the system's ability to identify food items, especially in scenarios with closely packed food images, thus improving accuracy. Validation results showcase the superior performance of the INMU iFood integrated V7-based system over its predecessor, V4-based, with notable improvements in protein and fat estimation. Furthermore, INMU iFood addresses limitations by offering users the option to import additional food products via a barcode scanner, thus providing access to a vast database of nutritional information through Open Food Facts. This integration ensures users can track their dietary intake effectively, with expanded access to over 3000 food items added to or updated in the Open Food Facts database covering a wide variety of dietary choices. Conclusions: INMU iFood is a promising tool for researchers, health care professionals, and individuals seeking to monitor their dietary intake within the context of Thai cuisine and for ultimately promoting better health outcomes and facilitating nutrition-related research.

Enhancing outcome prediction by applying the 2019 WHO DM classification to adults with hyperglycemic crises: A single-center cohort in Thailand.

BACKGROUND AND AIMS: Hyperglycemic crisis is a metabolic catastrophe which can occur in any type of diabetes. In 2019, the World Health Organization (WHO) revised the classification of diabetes mellitus (DM) and established two new hybrid forms, latent autoimmune diabetes in adults (LADA) and ketosis-prone type 2 diabetes (T2D). This study aimed to determine clinical outcomes after a hyperglycemic crisis event in people with diabetes classified subtypes by 2019 WHO DM classification. METHODS: A five-year (2015-2019) retrospective study of adult patients admitted with hyperglycemic crises was conducted. Types of diabetes were recategorized based on the 2019 WHO DM classification. Clinical characteristics, in-admission treatment and complications, long-term follow-up outcomes, and mortality were collected, analyzed, and compared. RESULTS: A total of 185 admissions occurred in 136 patients. The mean age was 50.6 ± 18.4 years (49.3 % men). The annual average incidence of hyperglycemic crises was 5.2 events/1000 persons. The proportion of type 1 diabetes, T2D, LADA, ketosis-prone T2D, and pancreatic DM were 15.4 %, 69.1 %, 2.2 %, 11 %, and 2.2 %, respectively. In-hospital mortality was 3.7 % while cumulative mortality totaled 19.1 %. During the 24-month follow-up, ketosis-prone T2D had the highest success of insulin discontinuation (HR 6.59; 95 % CI 6.69-319.4; p < 0.001), while T2D demonstrated the highest mortality compared to others (HR, 2.89; 95%CI 1.15-6.27; p = 0.02). CONCLUSION: The reclassification of diabetes based on 2019 WHO DM classification helped elucidate differences in long-term outcomes and mortality among DM types. The new classification, which separates ketosis-prone T2D from standard T2D, should be encouraged in clinical practice for precise and individualized management.

Real-World Use of Once-Weekly Semaglutide in Thai Patients With Type 2 Diabetes Mellitus in a Private Hospital Setting.

Objective: To evaluate the real-world use of once-weekly semaglutide among Thai patients with type 2 diabetes (T2DM) in a private hospital setting. Methodology: A retrospective review of Thai patients with T2DM who have initiated semaglutide for at least 1 month between June 2020 and March 2022 at Theptarin Hospital, Bangkok, Thailand. Results: A total of 58 patients (50% female, mean age 55.6 ± 15.9 years, with duration of diabetes 12.6 ± 10.3 years, BMI 31.5 ± 4.4 kg/m2, baseline HbA1c 7.9 ± 1.9%, with prior GLP-1 RA use 24.1%, and concomitant SGLT2i intake (41.4%) were included. During a median follow-up of 6 months, the mean serum HbA1c level reduction was 1.3 ± 1.7% with weight loss of 4.7 ± 4.1 kg. The proportion of patients who achieved optimal and sustainable glycemic control (HbA1c < 7.0%) increased from 43.1% to 55.8% at the last follow-up. The proportion of patients reaching both HbA1c targets of <7.0% and 5% weight loss was 27.8%. No cases of pancreatitis, cancer, or progressive retinopathy were observed. Conclusions: In this single center undertaking, it was shown that in among persons with T2DM and obesity in Thailand, semaglutide was associated with short-term glycemic control and weight loss comparable with what has been observed in randomized clinical trials and other RWE.

Implementation of diabetes care and educational program via telemedicine in patients with COVID-19 in home isolation in Thailand: A real-worldexperience.

BACKGROUND: The coronavirus disease (COVID-19) outbreak in Bangkok led to a shortage of hospital capacity, and a home isolation system was set up. We described the process of diabetes self-management education and support (DSMES) and glycemic management via telemedicine, along with outcomes in home-isolated patients with COVID-19 infection. METHODS: A retrospective chart review of glucose values, insulin and corticosteroids use, and outcomes was performed. RESULTS: A volunteer group of 21 endocrinologists and 21 diabetes educators/nurses formed the consultation team. Patients with diabetes or at high-risk of diabetes and receiving corticosteroids were referred by primary volunteer physicians. Glucometers and related supplies, and insulin were donated, and delivered via same-day delivery services. A chat group of an individual patient/their caregiver, diabetes educator, endocrinologist, and primary physician was formed (majority via LINE® platform) to assess the patient's clinical status and need. Real-time virtual DSMES sessions were performed and treatments were adjusted via smartphone application or telephone. There were 119 patients (1,398 service days), mean (SD) age 62.0 (13.6) years, 85.7% had a history of type 2 diabetes, and 84.0% received corticosteroids. Insulin was used in 88 patients; 69 of whom were insulin-naïve. During the first 10 days, there were 2,454 glucose values. The mean glucose level on day 1 was 280.6 (122.3) mg/dL, and declined to 167.7 (43.4) mg/dL on day 10. Hypoglycemia occurred in 1.4% of the values. A majority of patients (79.5%) recovered at home. CONCLUSION: Diabetes care and DSMES delivered via telemedicine to patients on home isolation during COVID-19 pandemic was safe and effective.

Celiac Disease as a Cause of Anemia and Brittle Diabetes in Type 1 Diabetes Mellitus.

Untreated celiac disease (CD) leads to an increased risk for hypoglycemia and diabetic complications. However, the diagnosis of CD can be challenging and some extra-gastrointestinal tract manifestations could be a presenting symptom. We report a case of a 29-year-old Indian male with brittle T1DM whose underlying CD was discovered from a work-up for anemia. After an introduction of a gluten-free diet, he gained 5 kgs in two months, was responsive to oral iron supplement, and had stable glycemic control with much less hypoglycemia. Even though this disease is rare in Asian populations, the diagnosis of celiac disease should always be kept in mind when people with T1DM present with unexplained microcytic anemia and/or unexplained hypoglycemia.

Feasibility Study of an Automated Carbohydrate Estimation System Using Thai Food Images in Comparison With Estimation by Dietitians.

Carbohydrate counting is essential for well-controlled blood glucose in people with type 1 diabetes, but to perform it precisely is challenging, especially for Thai foods. Consequently, we developed a deep learning-based system for automatic carbohydrate counting using Thai food images taken from smartphones. The newly constructed Thai food image dataset contained 256,178 ingredient objects with measured weight for 175 food categories among 75,232 images. These were used to train object detector and weight estimator algorithms. After training, the system had a Top-1 accuracy of 80.9% and a root mean square error (RMSE) for carbohydrate estimation of <10 g in the test dataset. Another set of 20 images, which contained 48 food items in total, was used to compare the accuracy of carbohydrate estimations between measured weight, system estimation, and eight experienced registered dietitians (RDs). System estimation error was 4%, while estimation errors from nearest, lowest, and highest carbohydrate among RDs were 0.7, 25.5, and 7.6%, respectively. The RMSE for carbohydrate estimations of the system and the lowest RD were 9.4 and 10.2, respectively. The system could perform with an estimation error of <10 g for 13/20 images, which placed it third behind only two of the best performing RDs: RD1 (15/20 images) and RD5 (14/20 images). Hence, the system was satisfactory in terms of accurately estimating carbohydrate content, with results being comparable with those of experienced dietitians.

Effectiveness of Insulin Degludec in Thai Patients with Diabetes Mellitus: Real-World Evidence From a Specialized Diabetes Center.

BACKGROUND: Insulin degludec, an ultra-long-acting insulin analogue, has been available in Thailand since October 2016. Although clinical trial results revealed less hypoglycemia, data from real-world settings is limited especially in Asian patients. This study aimed to evaluate prospectively the real-world effectiveness, safety, quality of life (QOL) and patient satisfaction with insulin degludec among Thai patients with diabetes mellitus (DM). METHODS: From October 2016 to September 2017, all patients who had started insulin degludec for at least 3 months were observed and evaluated at baseline, 3, 6, and 12 months. QOL was assessed using WHOQOL-BREF-THAI and level of satisfaction was measured by 7-point Likert scale. Glycemic fluctuation from paired iPro2 continuous glucose monitoring (CGM) obtained 4-6 weeks apart were also evaluated from a subset of patients with T1DM who switched from insulin glargine to insulin degludec. RESULTS: A total of 55 patients (T2DM 76.4%, females 54.5%, mean age 57.1±16.1 years, duration of diabetes 16.7±8.8 years, BMI 27.3±5.5 kg/m2, baseline A1C 9.3±2.3%, median duration of treatment 8 months) were included in the study. In T1DM patients (n=13), the overall mean A1C reduction at 12 months was 0.5% with minimal weight gain of 0.9 kgs at 12 months. In T2DM patients (n=42), the overall mean A1C reduction at 12 months was 0.8% with minimal weight loss of 0.4 kgs at 12 months. The proportion of T1DM patients who could achieve optimal glycemic control increased slightly from 14.3 to 18.2% but the proportion of T2DM patients who could achieve optimal glycemic control increased from 30.8 to 53.8%. Patient satisfaction showed a sustained improvement throughout the duration of study. In four T1DM patients who had paired CGM data, insulin degludec provided greater reductions in glycemic variability endpoints with increased time-in-range when compared with previous insulin glargine. DISCUSSION: Our data suggested that the effectiveness of insulin degludec was consistent with the results seen in clinical trials with lower risk of patients-reported hypoglycemia, and a significant improvement in glycemic control. Patients also reported higher treatment satisfaction. More long-term and cost-effectiveness data are needed to establish the role of this ultra-long-acting insulin in real-world settings.

Clinical characteristics and outcomes of care in adult patients with diabetic ketoacidosis: A retrospective study from a tertiary diabetes center in Thailand.

BACKGROUND: Diabetic ketoacidosis (DKA) is a metabolic catastrophe which could occur in any type of diabetes. Even when fundamental key points of DKA treatment had been followed, some differences exist in treatment protocols in each physician, highlighting the need to assess adherence to DKA guideline. AIM: This study aimed to examine trend of hospitalized DKA patients and outcomes of treatment over a decade at Theptarin Hospital, a multi-discipline based diabetes center in Thailand. METHOD: A retrospective study of DKA episodes admitted over a 14-year period (2005-2018) was done. Clinical characteristics, laboratory data, type of diabetes, severity of DKA were collected and analyzed. RESULTS: A total of 94 DKA episodes occurred in 81 diabetic patients (females 61.5%, mean age 47.4 ±â€¯20.4 years, T1DM 41.5%, T2DM 50.0%, Ketosis-prone diabetes 8.5%, baseline A1C 10.8 ±â€¯3.0%). While infection was the common precipitating factor in T2DM, omission of insulin was the usual precipitating factor in T1DM. During ongoing management, 26.6% of patients developed hypokalemia and supplementation was not prescribed as per protocol in this group of patients. Almost 13% of patients experienced hypoglycemia in the first 24 h. Median time to resolution of DKA was 8.5 h. Four T2DM patients expired from the precipitating cause of DKA which accounted for mortality rate at 4.3% in our study. CONCLUSIONS: Inadequate metabolic monitoring and iatrogenic hypoglycemia remain areas of concern for DKA management. Occurrence of hypokalemia was related to poor adherence to protocol guidance on potassium supplementation. A strengthened educational program for nursing and medical staffs should be emphasized.

Incidence and malignancy rates classified by The Bethesda System for Reporting Thyroid Cytopathology (TBSRTC) - An 8-year tertiary center experience in Thailand.

Incidence and Malignancy Rates Classified by The Bethesda System for Reporting Thyroid Cytopathology (TBSRTC) - An 8-year Tertiary Center experience in Thailand. BACKGROUND: Fine-needle aspiration (FNA) of the thyroid is considered the best diagnostic tool for preoperative evaluation of thyroid nodules. The introduction of The Bethesda System for Reporting Thyroid Cytopathology (TBSRTC) in 2010 provided the opportunity to establish worldwide standard for reporting and terminology guidelines for diagnostic categories. It is recommended that every institution evaluates the risk of malignancy (ROM) in each category for quality improvement process. AIM: To assess the effectiveness of TBSRTC method at our institution using cyto-histological correlation. METHOD: A retrospective 8-year (2010-2017) audit of thyroid FNA done by thyroid specialists at Theptarin hospital. The FNA results were classified according to TBSRTC. Histopathology reports for operated cases were used to correlate cytology and final histopathology. RESULTS: A total of 2735 thyroid FNA from 2115 patients (mean age 45.7 ±â€¯13.1 years, female 89.8%) were examined. The rates of non-diagnostic, benign, atypia of undetermined significance (AUS), follicular neoplasm, suspected for malignancy, and malignant cases were 21.1%, 66.6%, 4.7%, 2.4%, 1.8%, and 3.3% respectively. There were 188 patients (9%) who underwent surgical resection with available histopathology. Malignancy rates in operated thyroid nodules were 20.0%, 4.2%, 9.4%, 23.5%, 57.1%, and 90.3% for categories 1 to 6, respectively. The sensitivity, specificity, negative predictive value, and positive predictive value were 96.6%, 88.5%, 95.8%, and 90.3, respectively. CONCLUSIONS: Preoperative diagnosis of thyroid nodules using TBSRTC in our hospital was comparable with other studies. The uniform diagnostic criteria of the Bethesda System help avoid misinterpretation while sharing local experience with international benchmarks.

Performance of HbA1c versus oral glucose tolerance test (OGTT) as a screening tool to diagnose dysglycemic status in high-risk Thai patients.

BACKGROUND: Dysglycemic status defined by prediabetes and diabetes is known to be related with future risk of diabetic complications and cardiovascular diseases. Herein, we aimed to determine the diagnostic accuracy of glycated hemoglobin (HbA1c) when compared with oral glucose tolerance test (OGTT) as a reference test in identifying dysglycemic status among high-risk Thai patients receiving care in an out-patient setting. METHODS: An 11-year retrospective cross-sectional study of high-risk Thai patients who underwent OGTT during 2007-2017 was analysed. The OGTT was used as a reference test to identify subjects of dysglycemic status. The diagnostic accuracy of HbA1c and the agreement between HbA1c and OGTT were examined. Validated Thai diabetes risk score, Thai cardiovascular risk score (Thai CV risk score), and visceral fat area (VFA) were also compared in each glycemic status from OGTT as surrogate markers for future diabetes and cardiovascular diseases. RESULTS: A total of 512 subjects (females 60.5%, mean age of 50.3 ± 12.7 years, BMI of 26.5 ± 4.6 kg/m2) were reviewed. Normal glucose tolerance (NGT) was found in 220 patients (43.0%), impaired glucose tolerance (IGT) in 191 patients (37.3%), and diabetes in 101 patients (19.7%). The prevalence of diabetes using OGTT was approximately two times higher than those defined by HbA1c (19.7% versus 11.1%). There were poor agreements between the classifications of prediabetes and diabetes defined by OGTT and HbA1c (Cohen's Kappa 0.154 and 0.306, respectively). Using a cut-off value for HbA1c ≥6.5% as a threshold for HbA1c-defined criteria of diabetes, sensitivity was 32% (95% CI 23-41%) and specificity was 94% (95% CI 92-96%). The optimal cut-off HbA1c value for detecting diabetes by Youden's index was at HbA1c 6.2%. Thai CV risk score was much higher among the OGTT-defined diabetes group when compared with the NGT group (median score 10 vs. 3, p-value < 0.001). CONCLUSIONS: Despite the practicality and validity of HbA1c as a diagnostic test, our study suggested that HbA1c as a screening tool for diabetes in high-risk Thai patients is much inferior to OGTT. With limitations of HbA1c, physicians should continue to advocate OGTT as a screening tool for the identification of dysglycemic status in high-risk Thai patients.

Evaluation of Dietary Intakes and Nutritional Knowledge in Thai Patients with Type 2 Diabetes Mellitus.

INTRODUCTION: Most nutritional guidelines for diabetes management emphasize the importance of having individualized goals, away from a one-size-fits-all approach. However, there is a dearth of information on the dietary intakes and nutritional knowledge of Thai patients with type 2 diabetes mellitus (T2DM). This study is aimed at clarifying dietary intakes in relationship to glycemic control and at examining nutritional knowledge among Thai patients with T2DM. MATERIALS AND METHODS: A cross-sectional study of outpatients with T2DM at Theptarin Hospital and Ramathibodi Hospital (Bangkok, Thailand) was performed to assess dietary intakes by food records. Diabetes nutritional knowledge and dietary self-care behavior was also evaluated. RESULTS: A total of 304 Thai patients with T2DM (female 52.6%, mean age 57.4 ± 10.9 years, body mass index (BMI) 27.3 ± 4.8 kg/m2, and baseline A1C 7.2 ± 1.3%) participated in the study. The mean daily calorie intake was 1427 ± 425 kcal, and mean intake for each macronutrient was acceptable (carbohydrate 52%, protein 17%, and fat 31%). However, the intake of free sugar was much higher (12.1 ± 5.8% of total daily energy intake) and dietary fiber intake (9 grams per day) was much lower than recommended. There were no correlations between dietary intake and glycemic control. A subset of patients (N = 213) completed the diabetes nutritional knowledge survey. There was no association between diabetes nutritional knowledge and the actual dietary self-care behavior. CONCLUSION: These results indicate that compliance of Thai patients with T2DM to dietary recommendations is not completely satisfactory, especially for free sugar and dietary fiber intakes. Addressing the reality of how patients with T2DM eat in their daily lives and their knowledge gaps would enable them to adhere to medical nutrition therapy.

Effectiveness of long-term treatment with SGLT2 inhibitors: real-world evidence from a specialized diabetes center.

BACKGROUND: Diabetes is a progressive disease needing multiple drugs for achieving and maintaining good glycemic control. Sodium-glucose co-transporter 2 inhibitors (SGLT2i) is a novel class of anti-diabetic agent which offers several beneficial effects. However, the long-term effectiveness in clinical practice and safety data of SGLT2 inhibitors is limited, especially in Asian patients. To better understand the effectiveness of SGLT2i in clinical practice, we conducted a retrospective evaluation of patients with diabetes on SGLT2i. METHODS: This retrospective observational study uses data of patients with diabetes who had been prescribed SGLT2i and continued to use at least 6 months at Theptarin Hospital, Bangkok. The characteristics of patients, changes in glycemic control and body weight at 3, 6, 12, 18, 24 months and the last follow-up were evaluated. RESULTS: A total of 189 patients with diabetes (females 50.3%, mean age 59.9 ± 12.3 years, T2DM 97.3%, duration of diabetes 16.3 ± 9.2 years, baseline BMI 29.9 ± 6.1 kg/m2, baseline HbA1c 8.8 ± 1.6%) were prescribed SGLT2i during the study period. At the time of first SGLT2i prescription, 80.4% used three or more other anti-diabetic agents concomitantly and 34.6% used insulin concomitantly. 151 patients who continue to use at least 6 months were included in analysis. At the last follow-up (median time 16 months), overall median HbA1c reduction and weight reduction were 1.0% and 1.5 kg, respectively. While glycemic control could maintain up to 18 months, weight loss gradually rebounded after the first 6 months and then backed to baseline body weight at 18 months (78.2 ± 18.0 kg vs. 78.0 ± 17.8, p value = 0.324). The incidence of adverse drug reactions of special interest (polyuria, volume depletion-related events, urinary tract infection, genital infection, and hypoglycemia) was 2.1, 1.6, 2.1, 2.6, and 7.9%, respectively. DISCUSSION: This real-world study confirmed long-term durability of glycemic control with SGLT2i in not only monotherapy, but also add-on studies with other oral anti-diabetic drugs and/or insulin treatment. However, weight loss became evident early after 6 weeks then reached slightly rebounds after 24 weeks until the end of follow-up. Further studies should be done towards a better understanding of treatment with SGLT2i in routine clinical practice.

Development of New Lemon-Lime Flavored Beverage for OGTT: Acceptability and Reproducibility.

BACKGROUND: The oral glucose tolerance test (OGTT) is essential procedure in both screening and diagnosis of impaired glucose tolerance (IGT), impaired fasting glucose (IFG) and diabetes mellitus (DM), but it is not easy to perform because of intense sweetness of the 75-g glucose test beverage causing abdominal discomfort post-testing. Therefore, the new formula of non-carbonated lemon-lime flavored beverage was developed to increase its palatability and better compliance. OBJECTIVE: To develop a new non-carbonated lemon-lime flavored beverage to replace the standard beverage for OGTT Subsequently, the diagnostic value and acceptability between the new formula and the traditional 75-g OGTT formula were compared in healthy subjects. MATERIAL AND METHOD: The new lemon-lime flavored formula was developed to replace the standard beverage for OGTT by adding 1,000 milligram of citric acid and 0.03 gram of lime flavor to 75 gram of anhydrous glucose to a final volume of 300 ml. The study was conducted in 30 healthy subjects who underwent the traditional 75-gram OGTT test and the new formula of OGTT beverage one week later, or vice versa, to access acceptability, indices markers of insulin secretion, and insulin sensitivity. Palatability was determined by rating on a 9-point Hedonic Scale. RESULTS: Thirty healthy subjects (15 females) with the age of 33.2 ± 7.5 years and body mass index of 22.9 ± 3.5 kg/m² were enrolled. No significant difference was found between plasma glucose in 0, 30, 60, 90, and 120 minutes, insulin level (0 and 120 minutes) and four insulin surrogate markers in both traditional 75-gram OGTT and new formula of lemon-lime flavored OGTT beverage. The overall satisfaction score of the new formula OGTT was better when compared with the scores of the traditional OGTT (7.1 ± 1.8 vs. 4.7 ± 2.0). Only one subject complained about abdominal discomfort in both episode of OGTT CONCLUSION: The modified lemon-lime flavored beverage for OGTT demonstrated better acceptance in the subjects without difference in plasma glucose values and OGTT derived parameters responses to OGTT in comparison to the traditional formula.