Hasil untuk "Diseases of the endocrine glands. Clinical endocrinology"

E. Feldman, R. Nelson

Gabriela Rozo-Paz, C. Ruiz-Forero, José David Suárez-Mera et al.

Introduction Thyroid hormone resistance (THR) is a rare genetic syndrome characterized by reduced sensitivity to thyroid hormones. Patients may be asymptomatic, although clinical manifestations depend on the THR subtype. This entity commonly has abnormal thyroid function tests and can be confirmed by molecular analyses. Case presentation The present study describes a 55 year-old female diagnosed with surgically resected papillary thyroid carcinoma. During the endocrinology consults, elevated thyroid hormone levels were detected without an adequate TSH response, and THR was suspected. Moreover, Parkinson’s disease was diagnosed, and treatment with levodopa/carbidopa was initiated. Following this regimen, her TSH and total T3 levels were subsequently normalized, which suggests a potential effect of this agent on the normalization of these hormone levels in the blood. In this case, the role of levodopa was crucial to regulate the TSH concentration which was required to carry out the resection of a tumoral remnant. Conclusion The influence of dopamine in the endocrine system, specifically in the thyroid gland, is beneficial in conditions such as THR where abnormal TSH levels can be lowered, helping to balance the thyroid and hormones function.

Sisi Ma, Ryan Coopergard, Mark Clements et al.

Abstract BackgroundExercise is an important aspect of diabetes self-management. Patients with type 1 diabetes frequently struggle with exercise-induced hyperglycemia and hypoglycemia, decreasing their willingness to exercise. ObjectiveWe aim to build accurate and easy-to-deploy models to forecast exercise-induced glycemic events in real-world settings. MethodsWe analyzed free-living data from the Type 1 Diabetes Exercise Initiative study, where adults with type 1 diabetes wore a continuous glucose monitor (CGM) while performing video-guided exercises (30-minute exercises at least 6 times over 4 weeks), along with concurrent detailed phenotyping of their insulin program and diet. We built models to predict glycemic events (blood glucose ≤54 mg/dL, ≤70 mg/dL, ≥200 mg/dL, and ≥250 mg/dL) during and 1 hour post exercise with variables from 4 data modalities, such as demographic and clinical (eg, glycated hemoglobin; CGM (blood glucose value and their summary statistics); carbohydrate intake and insulin administration; and exercise type, duration, and intensity. We used repeated stratified nested cross-validation for model selection and performance estimation. We evaluated the relative contribution of the 4 input data modalities for predicting glycemic events, which informs the cost and benefit for including them in the decision support tool for risk prediction. We also evaluated other important aspects related to model translation into decision support tools, including model calibration and sensitivity to noisy inputs. ResultsOur models were built based on 1901 exercise episodes for 329 participants. The median age for the participants was 34 (IQR 26‐48) years. Of the participants, 74.8% (246/329) are female and 94.5% (329) are White. A total of 182/329 (55.3%) participants used a closed-loop insulin delivery system, while the rest used a pump without a closed-loop system. Models incorporating information from all 4 data modalities showed excellent predictive performance with cross-validated area under the receiver operating curves (AUROCs) ranging from mean 0.880 (SD 0.057) to mean 0.992 (SD 0.001) for different glycemic events. Models built with CGM data alone have statistically indistinguishable performance compared to models using all data modalities, indicating the other 3 data modalities do not add additional information with respect to predicting exercise-related glycemic events. The models based solely on CGM data also showed outstanding calibration (Brier score ≤0.08) and resilience to noisy input. ConclusionsWe successfully constructed models to forecast exercise-induced glycemic events using only CGM data as input with excellent predictive performance, calibration, and robustness. In addition, these models are based on automatically captured CGM data, thus easy to deploy and maintain and incurring minimal user burden, enabling model translation into a decision support tool.

Alyce S Adams, Catherine Lee, Gabriel Escobar et al.

Abstract BackgroundDiabetes-related lower extremity complications, such as foot ulceration and amputation, are on the rise, currently affecting nearly 131 million people worldwide. Methods for early detection of individuals at high risk remain elusive. While data-driven diabetic polyneuropathy algorithms exist, high-performing, clinically useful tools to assess risk are needed to improve clinical care. ObjectiveThis study aimed to develop an electronic medical record–based machine learning algorithm that would predict lower extremity complications. MethodsWe conducted a retrospective longitudinal cohort study to predict the risk of lower extremity complications within 24 months of an initial diagnosis of diabetic polyneuropathy. From an initial cohort of 468,162 individuals with at least 1 diagnosis of diabetic polyneuropathy at one of 2 multispecialty health care systems (based in northern California and Colorado) between April 2012 and December 2016, we created an analytic cohort of 48,209 adults with continuous enrollment, who were newly diagnosed with no evidence of end-of-life care. The outcome was any lower extremity complication, including foot ulceration, osteomyelitis, gangrene, or lower extremity amputation. We randomly split the data into training (38,569/48209; 80%) and testing (9,640/48209; 20%) datasets. In the training dataset, we used super Learner (SL), an ensemble learning method that employs cross-validation and combines multiple candidate risk predictors, into a single risk predictor. We evaluated the performance of the SL risk predictor in the testing dataset using the receiver operating characteristic curve and a calibration plot. ResultsOf the 48,209 individuals in the cohort, 2327 developed a lower extremity complication during follow-up. The SL risk estimator exhibited good discrimination (AUC=0.845, 95% CI 0.826-0.863) and calibration. A modified version of our SL algorithm, simplified to facilitate real-world adoption, had only slightly reduced discrimination (AUC=0.817, 95%CI 0.797-0.837). The modified version slightly outperformed the naïve logistic regression model (AUC=0.804, 95% CI 0.783-0.825) in terms of precision gained relative to the frequency of alerts and number of patients that needed to be evaluated. ConclusionsWe have built a machine learning–based risk estimator with the potential to improve clinical detection of diabetic patients at high risk for lower extremity complications at the time of an initial diabetic polyneuropathy diagnosis. The algorithm exhibited good discriminant validity and calibration using only data from the electronic medical record. Additional research will be needed to identify optimal contexts and strategies for maximizing algorithmic fairness in both interpretation and deployment.

Huimin Guo, Huimin Guo, Sen Wang et al.

ObjectiveThe triglyceride - glucose (TyG) index has been confirmed as an independent risk factor for ischemic stroke (IS) in numerous studies. In terms of the role of carotid ultrasound in the risk assessment of IS, the focus has shifted from merely concentrating on the degree of stenosis to paying more attention to the status of carotid plaques. However, there are limited studies on combining clinical indicators such as the TyG index with carotid ultrasound parameters to assess the risk of IS. Through a retrospective study, we aim to explore the role of combining these two types of indicators in the risk assessment of ISMethodsThis study included 145 patients with IS and 99 no ischemic stroke (NIS) patients diagnosed by magnetic resonance imaging (MRI) from January 2020 to June 2024. The TyG index was calculated as ln [fasting triglyceride (mg/dL) × fasting blood glucose (mg/dL)/2]. The carotid ultrasound parameters integrated were as follows: the presence or absence of carotid plaques, the location of the largest carotid plaque, carotid intima - media thickness (CIMT), the lengthness and thickness diameters of the largest carotid plaque, and the degree of carotid stenosis. Univariate (multivariate) logistic regression analysis, ROC curve analysis, etc. were conducted on the data using SPSS 26 and MATLAB Online. These were aimed at assessing the effectiveness of integrating clinical indicators with carotid ultrasound parameters in predicting the risk of IS.ResultsThe univariate logistic regression analysis (ULR) demonstrated that age, gender, TyG index, history of diabetes, history of hypertension, fasting blood glucose (FBG), systolic blood pressure(SBP), diastolic blood pressure(DBP), low-density lipoprotein cholesterol(LDL-C), cystatin C(Cys C), the presence or absence of carotid plaques, plaque location, carotid intima-media thickness(CIMT), the length and thickness of the largest plaque were significantly associated with IS (P < 0.05), while the P-values of triglycerides(TG), total cholesterol(TC), uric acid(UA) and carotid stenosis rate were greater than 0.05. The area under the ROC curve (AUC) of the TyG index for predicting IS was 0.645 (P < 0.001), indicating a certain predictive ability but relatively limited. The optimal cut-off value was 8.28, with a sensitivity of 0.83 and a specificity of 0.63 at this cut-off value. The stratified analysis based on quartiles of the TyG index revealed that as the TyG index increased, the prevalence of hypertension and diabetes, as well as multiple lipid and metabolic indicators, increased, and the characteristics of carotid plaques also changed. Multiple risk prediction models were constructed and analyzed by ROC curves. Model 1, which integrated traditional clinical indicators, TyG index and carotid ultrasound parameters, performed best (AUC = 0.932) (P < 0.001), while Model 16, which only included some carotid ultrasound indicators, had relatively low predictive efficacy (AUC = 0.750) (P < 0.001).ConclusionThis study confirms that the combination of TyG index and carotid ultrasound parameters is of great significance in predicting the risk of IS. The predictive ability of TyG index alone is limited, and Model 1 integrating multiple indicators has the best predictive effect and can provide a reference for clinical practice. However, due to the retrospective nature of this study and the limitations such as selection bias, small sample size and single-center, there are some discrepancies between some results and those of previous studies. Future studies need to conduct multi-center, large-sample studies and incorporate more factors to improve the model.

Yasuo Imanishi, Taku Furukubo, Shigeichi Shoji

Chronic kidney disease (CKD) induces secondary osteoporosis, characterized by an imbalance between bone formation and resorption due to kidney dysfunction; the result is a reduction in both bone mineral density and quality. This condition is compounded by disruption of bone metabolic turnover, abnormalities in bone microstructure and collagen cross-linking, and compromised bone quality, all of which contribute to increased bone fragility. Reduced kidney function is complicated by secondary hyperparathyroidism, which exacerbates bone fragility. Managing osteoporosis in patients with CKD is challenging because drugs may be contraindicated or require cautious administration, particularly those with high urinary excretion rates. In addition, severe hypercalcemia or hypocalcemia may develop in these patients following administration of active vitamin D or denosumab, respectively. The choice of pharmacotherapy depends on the stage of CKD; however, evidence for the safety and efficacy of osteoporosis drugs in moderate to severe cases of CKD, particularly stages G4, G5, and G5D (i.e., dialysis patients), is limited. This article focuses on the pathophysiology of CKD-associated osteoporosis, as well as the increased fracture risk, and provides a concise overview of safety considerations regarding administration of osteoporosis drugs in Japan. The data presented highlight the complexities associated with drug use in patients with CKD.

M. Santos, Laura Gago, Catarina Gama et al.

PV282 / #36Case Report Poster Topic:AS03 - Antiphospholipid SyndromeWe report a case of a 56-year-old male with a previous history of chronic mild thrombocytopenia, assumed to be a consequence of alcohol consumption (despite the absence of other alcoholic stigmas). He was admitted to the Internal Medicine Department with a 3-month history of extreme fatigue, anorexia, and weight loss (20%). Upon admission, he was hypotensive (100/54mmHg).Blood tests revealed normocytic normochromic anemia (Hb 9.9 g/dL), thrombocytopenia (94 000x10^6/L), elevated activated partial thromboplastin time (76.4s, N 28-40), hyperkaliemia (7.11 mmol/L, N 3.5-5.2) (but normal sodium) and elevated inflammatory parameters (CRP 2.25 mg/dL, ESR 118 mm/h). An extensive workup study was conducted to exclude malignancy and infection. A PET-FDG showed intense uptake in both adrenal glands, with heterogeneity and areas of necrosis, especially in the right adrenal gland (Figure 1A). The endocrinology department was consulted, and hormonal assessments revealed a low serum cortisol (1.9 μg/dL; N 6.2-19.4) and a high adrenocorticotropic hormone (626.0pg/ml; N 7.2-63.3). PAI was assumed and intravenous hydrocortisone (200 mg/day) was started, with subsequent clinical (blood pressure, constitutional symptoms) and laboratory (blood cells count and inflammatory markers) improvement. The main causes for PAI, namely autoimmune Addison’s disease, tuberculosis and human immunodeficiency virus infection, were excluded. At this moment, the Rheumatology department was consulted. Further workup revealed a positive lupus anticoagulant antibody (2 times in 12 weeks apart), ANAs 1/1280 (homogeneous nuclear pattern), anti-dsDNA antibodies elevation (517 UI/mL) and a weekly positive antinucleosome antibody. MRI scans showed atrophy of the adrenal glands (Figure 1B and 1C). The patient was diagnosed with SLE and APS, and after PAI control, he was discharged under glucocorticoid tapering (prednisolone 15 mg/day and fludrocortisone 0.05 mg/day), warfarin and hydroxychloroquine 400mg/day. Later on, azathioprine was also started (100 mg/day) and the patient remained asymptomatic and with normal laboratory parameters.Figure 1A: PET-FDG at diagnosis moment, showing intense uptake in both adrenal glands; B: abdominal MRI (T2) 3 months after the diagnosis, showing atrophy of both adrenal glands; C: abdominal MRI (T2) 6 months after the diagnosis, showing almost complete disappearance of both adrenal glands, comparing to the previous MRI.Antiphospholipid syndrome (APS) is a multisystemic autoimmune disorder characterized by recurrent arterial, venous and/or microvascular thrombotic events. The disease rarely affects the endocrine system, especially at presentation. The involvement of the adrenal gland, although rare, can be severe. Possible mechanisms behind adrenal manifestations include multiple microthrombosis of the suprarenal vein leading to infarction and adrenal hemorrhage, atrophy and finally failure (primary adrenal insufficiency [PAI]).[1][2]This case illustrates one of the rarest and still most severe consequences of APS. Patients with APS and adrenal hemorrhage, typically have bilateral involvement and develop adrenal insufficiency, just like our patient. [3] The disease can be fatal, thus early diagnosis and treatment as well as a close follow-up and multidisciplinary approach is needed to improve the prognosis of this rare disease.References:[1.] Hochberg MC. Chapter 135: Clinical features of systemic lupus erythematosus. In: Rheumatology 8th Ed. Elsevier;2022:P1113. [2.] Bouki K. Hormones (Athens) 2023;22(3): 521-31. [3.] Meade-Aguilar JA. Clin Immunol 2024;260:109906.

Hovra Zahoor, Shari Mitra, D. Lovre et al.

Abstract Disclosure: H. Zahoor: None. S. Mitra: None. D. Lovre: None. S. Gupta: None. BACKGROUND: Multiple endocrine neoplasia type 1 (MEN1) is a complex, autosomal dominant genetic disorder resulting from mutations in the MEN1 tumor suppressor gene which leads to tumors in the parathyroid, pituitary, and pancreas. We report the case of a 35-year-old male who presented with skeletal complaints of bilateral hip pain secondary to Brown tumors of the pelvis, ultimately leading to the diagnosis of MEN1. CASE PRESENTATION: A 35-year-old male with a past medical history of hypertension, prolactinoma with hypogonadism (diagnosed two years earlier) and gastroesophageal reflux disease was referred to the Endocrinology clinic by Orthopedic Oncology after a Computed Tomography (CT) of the abdomen and pelvis revealed findings consistent with a Brown tumor in the superior pubic ramus. A review of medical history revealed borderline hypercalcemia over the preceding several years and two episodes of nephrolithiasis. Further evaluation demonstrated a parathyroid hormone (PTH) level of 1220 pg/mL (15-65 pg/mL), calcium 10.6 mg/dL (8.4 - 10.4 mg/dL) with albumin 4.2 g/dL (3.4 - 5.0 g/dL), and elevated 24-hour urine calcium excretion of 478.4 mg/day (100.0 - 300.0 mg/day), confirming a diagnosis of primary hyperparathyroidism (PHPT). A sestamibi parathyroid scan indicated abnormal uptake in the left parathyroid gland and a follow-up 4DCT of the neck identified three abnormal lesions consistent with parathyroid adenomas. The patient underwent subtotal parathyroidectomy, and pathology confirmed hypercellular parathyroid tissue consistent with adenoma. The patient also had a history of prolactinoma with prior pituitary Magnetic Resonance Imaging (MRI) revealing a heterogeneous mass measuring 2 x 2.7 x 2 centimeters extending into the suprasellar region and abutting the terminal internal carotid arteries. At presentation, prolactin levels were 1719 ng/mL. He responded well to medical management with cabergoline 1.125mg twice a week. Surveillance CT imaging did not show any gastrointestinal or pancreatic tumors. Genetic testing identified a mutation in the MEN1 gene, confirming the diagnosis of MEN1 syndrome. DISCUSSION: Brown tumors are benign, fibrotic, erosive bony lesions caused by excessive osteoclastic activity in severe or longstanding hyperparathyroidism, typically in pelvic girdle, ribs, or clavicles. While PHPT is common in MEN1 syndrome, severe cases leading to Brown tumors are rare. Our case presents a unique clinical course in which hip pain and discovery of a Brown tumor revealed longstanding PHPT and led to the diagnosis of MEN1 syndrome. This case highlights the importance of comprehensive patient care, where thorough chart review and identification of abnormalities in unrelated imaging, along with incorporating a known diagnosis of prolactinoma help connect the dots to a rare diagnosis of MEN1 syndrome. Presentation: Sunday, July 13, 2025

Yueheng Jiang, Peng Zhang

Multi-disease diagnosis using multi-modal data like electronic health records and medical imaging is a critical clinical task. Although existing deep learning methods have achieved initial success in this area, a significant gap persists for their real-world application. This gap arises because they often overlook unavoidable practical challenges, such as modality missingness, noise, temporal asynchrony, and evidentiary inconsistency across modalities for different diseases. To overcome these limitations, we propose HGDC-Fuse, a novel framework that constructs a patient-centric multi-modal heterogeneous graph to robustly integrate asynchronous and incomplete multi-modal data. Moreover, we design a heterogeneous graph learning module to aggregate multi-source information, featuring a disease correlation-guided attention layer that resolves the modal inconsistency issue by learning disease-specific modality weights based on disease correlations. On the large-scale MIMIC-IV and MIMIC-CXR datasets, HGDC-Fuse significantly outperforms state-of-the-art methods.

David Restrepo, Ira Ktena, Maria Vakalopoulou et al.

Clinical decision-making relies on the integrated analysis of medical images and the associated clinical reports. While Vision-Language Models (VLMs) can offer a unified framework for such tasks, they can exhibit strong biases toward one modality, frequently overlooking critical visual cues in favor of textual information. In this work, we introduce Selective Modality Shifting (SMS), a perturbation-based approach to quantify a model's reliance on each modality in binary classification tasks. By systematically swapping images or text between samples with opposing labels, we expose modality-specific biases. We assess six open-source VLMs-four generalist models and two fine-tuned for medical data-on two medical imaging datasets with distinct modalities: MIMIC-CXR (chest X-ray) and FairVLMed (scanning laser ophthalmoscopy). By assessing model performance and the calibration of every model in both unperturbed and perturbed settings, we reveal a marked dependency on text input, which persists despite the presence of complementary visual information. We also perform a qualitative attention-based analysis which further confirms that image content is often overshadowed by text details. Our findings highlight the importance of designing and evaluating multimodal medical models that genuinely integrate visual and textual cues, rather than relying on single-modality signals.

William T. McBride, Mary Jo Kurth, Joanne Watt et al.

BackgroundIn patients undergoing orthopaedic trauma surgery, acute kidney injury (AKI) can develop post-operatively and is a major cause of increased mortality and hospital stay time. Development of AKI is associated with three main processes: inflammation, ischaemia-reperfusion injury (IRI) and hypoperfusion. In this study, we investigated whether ratios of urine and blood anti-inflammatory biomarkers and biomarkers of hypoperfusion, IRI and inflammation are elevated in patients who develop post-trauma orthopaedic surgery acute kidney injury (PTOS-AKI).MethodsBlood and urinary biomarkers of inflammation, hypoperfusion and IRI were analysed in 237 patients undergoing orthopaedic fracture surgery pre- and post-operatively. Biomarker ratios were compared between non-PTOS-AKI and PTOS-AKI patients.ResultsMultiple inflammatory biomarkers were significantly elevated in PTOS-AKI patients compared to non-PTOS-AKI patients. When urine anti-inflammatory biomarkers were expressed as biomarker ratios with biomarkers of inflammation, hypoperfusion and IRI, multiple ratios were lower in PTOS-AKI patients. In contrast, blood anti-inflammatory biomarkers when expressed as ratios with blood proinflammatory biomarkers were elevated in PTOS-AKI patients.DiscussionReductions in ratios of urine anti-inflammatory and proinflammatory biomarkers in PTOS-AKI patients suggest that the renal anti-inflammatory response is protective against the proinflammatory response in patients who do not develop PTOS-AKI. Detection of proinflammatory and anti-inflammatory biomarkers both pre- and post-operatively may be useful in detecting patients at risk of developing AKI after orthopaedic surgery.

A. R. Elfimova, A. Eremkina, O.Yu. Rebrova et al.

Background. Primary hyperparathyroidism (PHPT) is an endocrine disease characterized by excessive production of parathyroid hormone (PTH) and elevated or high-normal blood calcium levels caused by primary pathology of the parathyroid glands. The "classic" complication of PHPT is a decrease in the kidneys filtration function. Parathyroidectomy (PTE) reduces the risks of further deterioration in filtration function; however, in some cases, this is not achieved. Aim. To develop a mathematical model to predict the decline in estimated glomerular filtration rate (eGFR) 12 months after PTE in patients with PHPT, and implement it as a software. Methods. Retrospective study included 140 patients with PHPT who underwent PTE in 1993–2010 and 2018–2020 at the National Medical Research Center of Endocrinology. Analyzed variables included sex, age, indicators of calciumphosphorus, purine, lipid, and carbohydrate metabolism, presence of PHPT complications, treatment for PHPT, histological examination of removed parathyroid tissue, development of postoperative hypocalcemia and transient hypoparathyroidism, therapy for postoperative hypocalcemia. The random forest method was used to build the mathematical model. Results. To predict the decline in eGFR, a model using 24 predictors was built: sex, age, body mass index, PTH, ionized calcium, alkaline phosphatase, phosphorus, urea, eGFR, total cholesterol, diastolic blood pressure, SD(T-score)<-2.5/ SD(Z-score)<-2.0, CKD, duration of nephrolithiasis, use of angiotensin II receptor blockers and angiotensin-converting enzyme inhibitors, preoperative use of cholecalciferol and cinacalcet, parathyroid hyperplasia/adenoma, postoperative hypocalcemia, dose of alfacalcidol and calcium supplements, postoperative use of cholecalciferol. The resulting model (http://194.87.111.169/cfr) predicts a decline in eGFR in patients with PHPT after PTE with a probability of 56.8–86.3% and excludes – with a probability of 85.6–97.7%. Conclusion. A mathematical model to predict the decline in eGFR 12 months after PTE in patients with PHPT was developed, with an overall accuracy of 88%, 95% CI (79%; 93%). The model was implemented as a calculator that can be used in routine clinical practice.

Seda Karslı, M. Mert, Sema Çiftçi et al.

Abstract Introduction Multiple endocrine neoplasia 1 (MEN 1) is a rare autosomal dominantly inherited syndrome that results from mutations in the MEN1 gene. It is characterized by multiple tumors of endocrine glands and some other tissues. Early diagnosis is crucial for optimal management. Here we present a patient with delayed diagnosis and preventable metastatic disease. Clinical Case A 46-year-old male patient who was referred from gastroenterology clinic to our outpatient clinic following distal pancreatectomy. The postoperative pathology was reported as a well-differentiated neuroendocrine tumor (NET) with 3 focuses (60 mm, 9 mm and 2.5 mm in size). Six months ago, he underwent to Magnetic Resonance Imagination (MRI) of the abdomen because of abdominal pain and 17 kg weight loss in a year, that revealed a 69×67 mm cystic lesion in the tail of the pancreas (Figure 1A). Laboratory and physical examinations were in accordance with a nonfunctioning lesion (Table 1). The findings of the Gallium-68 DOTATATE PET/CT suggested residual and metastatic disease during follow-up (At the level of the pancreatic corpus, increased uptake was detected in two lesion sites adjacent to liver segment 3 and in various nodular lesion areas, the largest of which was approximately 20 mm in diameter, adjacent to the duodenum and pancreas (Figure 1B and 1D). No increased uptake was observed in the pancreas and these lesions in preoperative FDG/PET (Figure is not shown). The patient was consultated by oncology clinic and 177-Lutetium-DOTATATE and octreotide treatments were started. Parathyroidectomy and total thyroidectomy 7 years ago was noted in the patient's medical history. Persisting hypercalcemia and elevated PTH were also remarkable in laboratory testing (Table 2). The patient didn't apply to endocrinology clinic after parathyroidectomy and hypercalcemia had been overlooked for 7 years. Further investigation revealed a nonfunctioning 4 mm diameter pituitary lesion (Figure 1C). So the diagnosis of multiple endocrine neoplasia type 1 was suspected. Genetic test detected a c.784-2A>G spice site variant in MEN1 gene in heterozygous form. The patient denied a family history for MEN 1 but he reported that his father died at the age of 40 and his uncle at the age of 39 due to an illness of unknown origin, and two sons of the same uncle were followed up for parathyroid disease. Family screening and genetic counseling were recommended. Surgery of the remaining parathyroid glands and implantation in the forearm and thymectomy were planned. Conclusion MEN syndromes are rare and hypercalcemia is generally the initial presentation. Persistence of hypercalcemia following parathyroidectomy should be warning, and the possibility of MEN -1 syndrome should also be kept in mind even in the lack of family history. As in the presented case, diagnosis may be delayed due to late appearance of symptoms of other components, especially nonfunctional pancreaticoduodenal tumors.Figure 1. MRI and Ga-68 PET/CT images of the patient A: A 69×67 mm cystic lesion in the tail of the pancreas B: Increased Gallium-68 DOTATATE uptake at the level of the pancreatic corpus adjacent to liver segment 3 (SUVmax: 67.90) C: A nonfunctioning 4 mm diameter lesion in the right half of the pituitary D: Increased uptake of Gallium-68 DOTATATE in nodular lesions adjacent to the duodenum and pancreas (SUVmax: 31.09) Table 1. Laboratory results of the patient ACTH: Adrenocorticotropic hormone, FSH: follicle stimulating hormone, LH: luteinizing hormone, IGF-1: Insülin-like growth factor, TSH: Thyroid stimulating hormone, 1 mg DST: dexamethasone suppression test Table 2. Calcium and PTH levels of the patient before parathyroidectomy and following parathyroidectomy *Before parathyroidectomy ** Following parathyroidectomy PTH: Parathyroid hormone

Ziyan Yao, Fei Lin, Sheng Chai et al.

In this paper, an innovative multi-modal deep learning model is proposed to deeply integrate heterogeneous information from medical images and clinical reports. First, for medical images, convolutional neural networks were used to extract high-dimensional features and capture key visual information such as focal details, texture and spatial distribution. Secondly, for clinical report text, a two-way long and short-term memory network combined with an attention mechanism is used for deep semantic understanding, and key statements related to the disease are accurately captured. The two features interact and integrate effectively through the designed multi-modal fusion layer to realize the joint representation learning of image and text. In the empirical study, we selected a large medical image database covering a variety of diseases, combined with corresponding clinical reports for model training and validation. The proposed multimodal deep learning model demonstrated substantial superiority in the realms of disease classification, lesion localization, and clinical description generation, as evidenced by the experimental results.

Thomas A. Lasko, John M. Still, Thomas Z. Li et al.

Insufficiently precise diagnosis of clinical disease is likely responsible for many treatment failures, even for common conditions and treatments. With a large enough dataset, it may be possible to use unsupervised machine learning to define clinical disease patterns more precisely. We present an approach to learning these patterns by using probabilistic independence to disentangle the imprint on the medical record of causal latent sources of disease. We inferred a broad set of 2000 clinical signatures of latent sources from 9195 variables in 269,099 Electronic Health Records. The learned signatures produced better discrimination than the original variables in a lung cancer prediction task unknown to the inference algorithm, predicting 3-year malignancy in patients with no history of cancer before a solitary lung nodule was discovered. More importantly, the signatures' greater explanatory power identified pre-nodule signatures of apparently undiagnosed cancer in many of those patients.

Satya Kesav Gundabathula, Sriram R Kolar

This paper describes our approach to the MEDIQA-CORR shared task, which involves error detection and correction in clinical notes curated by medical professionals. This task involves handling three subtasks: detecting the presence of errors, identifying the specific sentence containing the error, and correcting it. Through our work, we aim to assess the capabilities of Large Language Models (LLMs) trained on a vast corpora of internet data that contain both factual and unreliable information. We propose to comprehensively address all subtasks together, and suggest employing a unique prompt-based in-context learning strategy. We will evaluate its efficacy in this specialized task demanding a combination of general reasoning and medical knowledge. In medical systems where prediction errors can have grave consequences, we propose leveraging self-consistency and ensemble methods to enhance error correction and error detection performance.

Dirk Douwes-Schultz, Alexandra M. Schmidt, Laís Picinini Freitas et al.

Univariate zero-inflated models are increasingly being used to account for excess zeros in spatio-temporal infectious disease counts. However, the multivariate case is challenging due to the need to account for correlations across space, time and disease in both the count and zero-inflated components of the model. We are interested in comparing the transmission dynamics of several co-circulating infectious diseases across space and time, where some of the diseases can be absent for long periods. We first assume there is a baseline disease that is well-established and always present in the region. The other diseases switch between periods of presence and absence in each area through a series of coupled Markov chains, which account for long periods of disease absence, disease interactions and disease spread from neighboring areas. Since we are mainly interested in comparing the diseases, we assume the cases of the present diseases in an area jointly follow an autoregressive multinomial model. We use the multinomial model to investigate whether there are associations between certain factors, such as temperature, and differences in the transmission intensity of the diseases. Inference is performed using efficient Bayesian Markov chain Monte Carlo methods based on jointly sampling all unknown presence indicators. We apply the model to spatio-temporal counts of dengue, Zika, and chikungunya cases in Rio de Janeiro, during the first triple epidemic there.

Adrian Lindenmeyer, Malte Blattmann, Stefan Franke et al.

Widespread adoption of AI for medical decision making is still hindered due to ethical and safety-related concerns. For AI-based decision support systems in healthcare settings it is paramount to be reliable and trustworthy. Common deep learning approaches, however, have the tendency towards overconfidence under data shift. Such inappropriate extrapolation beyond evidence-based scenarios may have dire consequences. This highlights the importance of reliable estimation of local uncertainty and its communication to the end user. While stochastic neural networks have been heralded as a potential solution to these issues, this study investigates their actual reliability in clinical applications. We centered our analysis on the exemplary use case of mortality prediction for ICU hospitalizations using EHR from MIMIC3 study. For predictions on the EHR time series, Encoder-Only Transformer models were employed. Stochasticity of model functions was achieved by incorporating common methods such as Bayesian neural network layers and model ensembles. Our models achieve state of the art performance in terms of discrimination performance (AUC ROC: 0.868+-0.011, AUC PR: 0.554+-0.034) and calibration on the mortality prediction benchmark. However, epistemic uncertainty is critically underestimated by the selected stochastic deep learning methods. A heuristic proof for the responsible collapse of the posterior distribution is provided. Our findings reveal the inadequacy of commonly used stochastic deep learning approaches to reliably recognize OoD samples. In both methods, unsubstantiated model confidence is not prevented due to strongly biased functional posteriors, rendering them inappropriate for reliable clinical decision support. This highlights the need for approaches with more strictly enforced or inherent distance-awareness to known data points, e.g., using kernel-based techniques.

Carlotta Cocchetti, Dominik Rachoń, Alessandra D. Fisher

Afshan Masood, Abeer Malkawi, Mohamed Siaj et al.