Low Positive Predictive Value Research Articles

Refining the 30-day emergency department visit after initial chemotherapy administration A-PROACCT model.

e23253 Background: This study attempted to address limitations in the A-PROACCT model developed by Stein et al. (JOP, 2023) to predict the rate of acute care events within 30 days (ACE30) of initial chemotherapy administration (ICA). The four most important predictors were drug category (immunotherapy), cancer type (gastrointestinal), age group ( < 40), and BMI category (underweight). Only linear relationships were considered. For patients for whom an ACE30 was predicted by their first and second models, only 0.18 and 0.23 actually had an ACE30, a low positive predictive value (PPV). We attempted to improve the PPV with advanced modeling techniques. Methods: Using Orlando Health data between February of 2012 and April 2021, random sampling was used to split the dataset into training (10,838) and validation (4645). For the training data, resampling was employed to obtain a balanced quota, equal observations followed by an ACE30 and not followed by an ACE30. We included only variables used in Stein (drug category, age group, ED visits and hospitalizations in the prior year, cancer type, insurance category, number of anti-cancer agents, race, and BMI category). Logistic regression with L1-penalty, XGBoost (a non-linear nonparametric tree-based algorithm to account for nonlinear relationships), and artificial neural networks were used to develop and evaluate predictive models. Different sampling methods (bootstrap and SMOTE) as well as cuff-off thresholds for high-risk groups were tested. Results: Based on evaluation with validation data set, the best performing approach was an XGBoost model with SMOTE resampling in the training data. The four most important predictors were ED visits in the prior year, payor category (self-insured), cancer type (genitourinary) and hospitalization in past year. This model reported a PPV of 0.27. Many other combinations of methodologies described above were performed, and the PPV varied between 0.19 to 0.27. Of the 282 ICAs identified as high-risk by the best model, 76 (27.0%) had an ACE30. Of the 4346 ICAs identified as low-risk, 533 (12.3%) had an ACE30. The differences between high risk and low risk ACE30 rates were statistically significant (p < 0.0001). In comparison, using Stein et al. models with our data reported a PPV of 0.27 and 0.25 which are basically at the same level of our best trained model. Conclusions: Using the variables included in A-PROACCT and a variety of machine learning models, advanced sampling methodologies and threshold cut-off limits, the best results were similar to those obtained using basic logistic regression. This suggests improvement in predicting acute care events following chemotherapy administration need to incorporate more extensive multimodal data, such as vital signs, performance status, patient reported outcomes, socioeconomic factors, laboratory and radiology results, remote patient monitoring and wearable device data.

Diagnosis and non-invasive assessment of MASLD in type 2 diabetes and obesity.

Metabolic dysfunction-associated steatotic liver disease (MASLD) is currently the most common chronic liver disease and an important cause of cirrhosis and hepatocellular carcinoma. It is strongly associated with type 2 diabetes and obesity. Because of the huge number of patients at risk of MASLD, it is imperative to use non-invasive tests appropriately. To provide a narrative review on the performance and limitations of non-invasive tests, with a special emphasis on the impact of diabetes and obesity. We searched PubMed and Cochrane databases for articles published from 1990 to August 2023. Abdominal ultrasonography remains the primary method to diagnose hepatic steatosis, while magnetic resonance imaging proton density fat fraction is currently the gold standard to quantify steatosis. Simple fibrosis scores such as the Fibrosis-4 index are well suited as initial assessment in primary care and non-hepatology settings to rule out advanced fibrosis and future risk of liver-related complications. However, because of its low positive predictive value, an abnormal test should be followed by specific blood (e.g. Enhanced Liver Fibrosis score) or imaging biomarkers (e.g. vibration-controlled transient elastography and magnetic resonance elastography) of fibrosis. Some non-invasive tests of fibrosis appear to be less accurate in patients with diabetes. Obesity also affects the performance of abdominal ultrasonography and transient elastography, whereas magnetic resonance imaging may not be feasible in some patients with severe obesity. This article highlights issues surrounding the clinical application of non-invasive tests for MASLD in patients with type 2 diabetes and obesity.

Validity of obstructive sleep apnoea diagnoses

INTRODUCTION. Healthcare databases are a valuable source for epidemiological research in obstructive sleep apnoea, but accurately registered diagnoses are pivotal in contributing quality evidence. We examined positive predictive values (PPV) of the International Classification of Diseases, tenth version (ICD-10) diagnosis for “obstructive sleep apnoea” and “sleep apnoea” in the Danish National Patient Register. METHODS. Using the Danish National Patient Registry, we randomly sampled 100 patients from the North Denmark Region diagnosed with “obstructive sleep apnoea” (ICD-10 code DG4732) and 100 patients diagnosed with “sleep apnoea” (DG473*) during the year 2020. We calculated the PPV using a documented Apnea-Hypopnea Index (AHI) ≥ 5 to confirm the recorded diagnosis. A total of 70 patients were referred to the private sector for assessment of the AHI and excluded due to limited access to their data. RESULTS. The study population included 130 patients, among whom 64 were diagnosed with “obstructive sleep apnoea”, and 66 patients were registered with “sleep apnoea”. The PPV for “obstructive sleep apnoea” was 93.8% (95% confidence interval (CI): 85.0-97.5%), and the PPV for “sleep apnoea” was 80.3% (95% CI: 69.2-88.1%). CONCLUSIONS. Our findings indicated a high validity of the ICD-10 code DG4732 with a PPV of 93.8% and a lower PPV (80.3%) for the ICD-10 code DG473* for identifying patients with obstructive sleep. The “obstructive sleep apnoea” diagnosis is a suitable source of data for epidemiological research to identify patients with the disease. FUNDING. None. TRIAL REGISTRATION. Not relevant.

Local vaginal bioelectrical impedance can predict preterm delivery in mice.

Preterm birth is a serious pregnancy complication that affects neonatal mortality, morbidity, and long-term neurological prognosis. Predicting spontaneous preterm delivery (PTD) is important for its management. While excluding the risk of PTD is important, identifying women at high risk of PTD is imperative for medical intervention. Currently used PTD prediction parameters in clinical practice have shown high negative predictive values, but low positive predictive values. We focused on sulfated and sialylated glycocalyx changes in the uterus and vagina prior to the onset of parturition and explored the potential of electrophysiological detection of these changes as a PTD prediction parameter with a high positive predictive value. In vivo local vaginal bioelectrical impedance (VZ) was measured using two different mouse PTD models. PTD was induced in ICR mice through the subcutaneous injection of mifepristone or local intrauterine injection of lipopolysaccharide (LPS). The PTD rates were 100% and 60% post-administration of mifepristone (16-20 h, n = 4) and LPS (12-24 h, n = 20), respectively. The local VZ values (15 and 10 h after mifepristone or LPS treatment, respectively) were significantly lower in the PTD group than in the non-PTD group. Receiver operator characteristic (ROC) curve analysis of VZ at 125 kHz as a predictor of PTD showed an area under the ROC curve of 1.00 and 0.77 and positive predictive values of 1.00 and 0.86, for the mifepristone and LPS models, respectively, suggesting that local VZ value can predict PTD. Histological examination of the LPS-treated model 6 h post-treatment revealed increased expression of sulfomucins and/or sulfated proteoglycans and sialomucins in the cervical epithelium, cervical stroma and vaginal stroma. In conclusion, local VZ values can determine sulfated and sialylated glycocalyx alterations within the uterus and vagina and might be a useful PTD prediction parameter.

Validity of self-reported hysterectomy and oophorectomy in a population-based cohort: The HUNT study.

To validate self-reported hysterectomy and bilateral oophorectomy. Validation study. Large population-based cohort study in Norway: The Trøndelag Health Study (HUNT). The Trøndelag Health Study 2 and 3 (HUNT2 and HUNT3) included questions on gynaecological history. Women who answered questions regarding hysterectomy and/or oophorectomy were included. In total, 30 263 women were included from HUNT2 (1995-1997) and 23 138 from HUNT3 (2006-2008), of which 16 261 attended both HUNT2 and HUNT3. We compared self-reported hysterectomy and bilateral oophorectomy with electronic hospital procedure codes. Sensitivity, specificity, positive predictive value and negative predictive value of self-reported hysterectomy and bilateral oophorectomy, by comparing with hospital procedure codes. Self-reported hysterectomy and bilateral oophorectomy in HUNT2 and/or HUNT3 both had specificity and negative predictive value above 99%. Self-reported hysterectomy had a sensitivity of 95.9%, and for bilateral oophorectomy sensitivity was 91.2%. Positive predictive value of self-reported hysterectomy was 85.8%, but for self-reported bilateral oophorectomy it was 65.4%. Self-reported hysterectomy corresponded quite well with hospital data and can be used in epidemiological studies. Self-reported bilateral oophorectomy, on the other hand, had low positive predictive value, and results based on such data should be interpreted with caution. Women who report no previous hysterectomy or bilateral oophorectomy can safely be classified as unexposed to these surgeries.

Cross-Cultural Adaptation and Validation of the Portuguese Version of the SARC-F in Community-Dwelling Older Adults.

(1) Background: The goal of this study was to analyze the reliability and validity of the Portuguese version of the SARC-F in older adults. (2) Methods: A total of 100 participants (77.1 ± 7.36 years, 73% women) were included in the study. In a first phase, the Portuguese SARC-F was adapted following the standardized forward-backward translation procedure, and internal consistency as well as inter-rater and test-retest reliability of the Portuguese SARC-F were analyzed. Secondly, clinical validation was evaluated by comparing the SARC-F total score with five operational definitions of sarcopenia and with other sarcopenia-related measurements. Discriminant validity, with respect to low muscle mass and strength and physical function were analyzed. (3) Results: The Portuguese SAR-F showed acceptable internal consistency (Cronbach α = 0.82), excellent inter-rater reliability (total score), and substantial to excellent test-retest reliability (ICC = 0.891 for the total score). Specificity ranged from 72.5% (FNIH) to 73.4 (IGWS), and negative predictive values went from 91.8% (EWGSOP1) to 97.3% (FNIH), but low sensitivity and positive predictive value were observed. The Portuguese SARC-F showed a moderate ability to discriminate people with low muscle strength (AUC = 0.78) and gait speed (AUC = 0.89). (4) Conclusions: The Portuguese SARC-F is a valid and reliable tool for ruling out sarcopenia in community-dwelling older adults and can discriminate between people with low handgrip strength and gait speed.

Validation of an ICD-Code-Based Case Definition for Psychotic Illness Across Three Health Systems.

Psychosis-associated diagnostic codes are increasingly being utilized as case definitions for electronic health record (EHR)-based algorithms to predict and detect psychosis. However, data on the validity of psychosis-related diagnostic codes is limited. We evaluated the positive predictive value (PPV) of International Classification of Diseases (ICD) codes for psychosis. Using EHRs at 3 health systems, ICD codes comprising primary psychotic disorders and mood disorders with psychosis were grouped into 5 higher-order groups. 1133 records were sampled for chart review using the full EHR. PPVs (the probability of chart-confirmed psychosis given ICD psychosis codes) were calculated across multiple treatment settings. PPVs across all diagnostic groups and hospital systems exceeded 70%: Mass General Brigham 0.72 [95% CI 0.68-0.77], Boston Children's Hospital 0.80 [0.75-0.84], and Boston Medical Center 0.83 [0.79-0.86]. Schizoaffective disorder PPVs were consistently the highest across sites (0.80-0.92) and major depressive disorder with psychosis were the most variable (0.57-0.79). To determine if the first documented code captured first-episode psychosis (FEP), we excluded cases with prior chart evidence of a diagnosis of or treatment for a psychotic illness, yielding substantially lower PPVs (0.08-0.62). We found that the first documented psychosis diagnostic code accurately captured true episodes of psychosis but was a poor index of FEP. These data have important implications for the case definitions used in the development of risk prediction models designed to predict or detect undiagnosed psychosis.

Optical imaging (HandScan) can identify ultrasound remission in rheumatoid arthritis

BackgroundIdentifying remission is of high importance in rheumatoid arthritis (RA) because remission is associated with less structural progression. We investigated the efficacy of a new optical imaging device, HandScan, to identify RA remission, as defined by ultrasound (US).Methods61 RA patients were included. Disease activity was evaluated by clinical assessment and US, using gray-scale (GS) and Power Doppler (PD). HandScan determined unitary optical spectral transmission (OST) values for wrists, metacarpophalangeal and proximal interphalangeal joints. At the patient level, three composite HandScan (HS) scores were calculated: total HS score; disease activity score OST (DAS-OST) and DAS-OST without patient global assessment (PtGA). Using ROC curves, we determined HS cut-offs to identify US-defined remission.ResultsAt the joint level, unitary OST values significantly correlated with GS synovitis [odds ratio (OR) 2.43, p < 0.0001] and PD positivity (OR 3.72, p = 0.0002 ). At the patient level, total HS score and DAS-OST were significantly associated with all gray-scale US (GSUS) and power doppler US (PDUS) parameters evaluated (synovitis number and grade, synovial thickness, PD grade) (p < 0.05). The cut-off to identify US-defined remission at the joint level was of 0.92, giving an 81% sensitivity and a 96% positive predictive value (PPV). At the patient level, ROC-curves failed to identify a robust cut-off for the total HS score, but did identify a cut-off (3.68) for DAS-OST to identify US-defined remission, but with lower sensitivity (75%), specificity (56%) and PPV (67%).ConclusionsHandScan is a non-invasive optical imaging technique providing OST values that correlate with GSUS and PDUS parameters. In addition, HandScan is able to reliably identify US-defined remission in RA at the joint level, with a good sensitivity and high PPV. At the patient level, HandScan DAS-OST can also determine US remission (while total HS score failed to do so), but with lower performance.

Abstract PO3-03-03: Prospective non-randomized study to compare accuracy of clinical examination under anesthesia, axillary ultrasound and histo-pathological evaluation for axillary nodal staging in women with clinically N0 early breast cancer

Abstract Introduction: Accurate assessment of axillary lymph nodes is crucial in the management of early breast cancer (EBC), especially in clinically node negative (cN0) axilla to avoid extensive axillary surgery. Clinical examination alone underestimates nodal disease in nearly 30% women with cN0 axilla. The current study compares, in cN0 axilla, the benefit of axillary ultrasonography (USG) and clinical axillary examination under anesthesia (EUA) to predict involvement of axillary lymph nodes. The gold standard in these patients however remains pathological evaluation after sentinel node biopsy/low axillary sampling (SNB/LAS) and a complete axillary lymph node dissection (ALND) if node positive. Methodology: Prospectively, 500 women with cN0 EBC were enrolled from Aug 2015 to April 2023 in a study approved by Institutional Ethics Committee. After informed consenting, a preoperative axillary USG was carried out in addition to standard breast imaging to determine number of axillary node(s) and its architecture. The USG assessment was labeled as suspicious or not and the result was blinded to the surgeons. A USG-guided FNAC was not performed as it would then be difficult to blind the surgeon and pathologist preoperatively. During surgery, an initial axillary EUA was performed before starting and any suspicious node was documented. This was followed by axillary staging by standard dual tracer SNB/LAS. A complete axillary dissection was done (level 1-3) if any node was positive on frozen section evaluation or final histopathology. Axillary node histopathology was the gold standard for comparison of effectiveness of clinical exam, USG, EUA, and SNB/LAS for prediction of axilla. Standard diagnostic tests such as sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV) and accuracy were used. Results: Thirty-six patients were excluded in the final analysis (disease progression, chemotherapy first, or had a surgery elsewhere). Of the eligible 464 cN0 patients, 129 were detected to have axillary metastases (27.8%) in final histopathology. The 2 interventions namely USG axilla, EUA were compared to final axillary nodal histopathology. Axillary USG reported suspicious/indeterminate node(s) in 129 (27.8%) patients. USG had a low sensitivity of 46.5% and a low PPV of 46.5% to identify a positive node. However, the specificity and NPV both were 79.4%. Axillary USG was 70.2% accurate in predicting axillary nodal involvement. EUA also had sensitivity of 60% and low PPV of 14.8%. However, the specificity of EUA was 73.4%, NPV of 95.9%; higher than that of USG. EUA was 72.4% accurate. SNB/LAS had the sensitivity of 93.3%, specificity 79.2%, NPV 82.9%, PPV 91.6% and accuracy rate 89.2% in predicting a positive axilla. Conclusions: While the fallacy of clinical exam remains at 27.8%, both USG alone (without FNAC) and EUA failed in predicting a positive axillary node. EUA fared better at predicting a negative axilla. USG guided FNAC would perhaps improve the sensitivity of USG, however additional investigations are difficult in resource constraint and high-volume center, especially, when surgical interventions like SNB or LAS remain standard of care. Table 1 Citation Format: Vani Parmar, Zeal Sanghvi, Shalaka Joshi, Nita Nair, Palak Popat, Seema Kembhavi, Souwmyashree KN, Soujanya Mynalli, Purvi Thakkar, Garvit Chitkara, Sangeeta Desai, Tanuja Shet, Rajendra Badwe. Prospective non-randomized study to compare accuracy of clinical examination under anesthesia, axillary ultrasound and histo-pathological evaluation for axillary nodal staging in women with clinically N0 early breast cancer [abstract]. In: Proceedings of the 2023 San Antonio Breast Cancer Symposium; 2023 Dec 5-9; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2024;84(9 Suppl):Abstract nr PO3-03-03.

Coffee ground emesis and early endoscopy: Is timing actually relevant?

ObjectiveCoffee ground vomiting is a classical symptom of upper gastrointestinal bleeding. However, the clinical usefulness is limited, due to the low positive predictive value. Our goal is to determine if whether an urgent endoscopy does modify our therapeutic management with a real impact on survival. Patients and methodsIt is a retrospective, observational and descriptive study. We selected all patients that underwent a gastroscopy in our center for coffee ground vomiting over the last 4 years (2017–2021). Two groups were established: urgent endoscopy (first 24 h) and scheduled (over 24 h). Then we studied differences between both groups regarding survival, ICU admission, hospitalization days and rebleeding. Results314 patients were identified, from which 276 were included, with 176 belonging to the urgency group and 109 to the scheduled group. There were no differences in the ICU admission, hospitalization days, survival or rebleeding after 30 days. There were no differences either in the number of potentially bleeding lesions nor the need of endoscopic therapeutic. ConclusionsCoffee ground vomiting, without any other data supporting upper gastrointestinal bleeding, does not represent a reliable indicator. Performing urgent endoscopy is not beneficial in terms of morbimortality. Therefore, a more conservative strategy would allow to differ endoscopy, decreasing risks and reducing costs, without affecting the prognosis.

Use of pressure muscle index to predict the contribution of patient's inspiratory effort during pressure support ventilation: a prospective physiological study.

The successful implementation of assisted ventilation depends on matching the patient's effort with the ventilator support. Pressure muscle index (PMI), an airway pressure based measurement, has been used as noninvasive monitoring to assess the patient's inspiratory effort. The authors aimed to evaluate the feasibility of pressure support adjustment according to the PMI target and the diagnostic performance of PMI to predict the contribution of the patient's effort during ventilator support. In this prospective physiological study, 22 adult patients undergoing pressure support ventilation were enrolled. After an end-inspiratory airway occlusion, airway pressure reached a plateau, and the magnitude of change in plateau from peak airway pressure was defined as PMI. Pressure support was adjusted to obtain the PMI which was closest to -1, 0, +1, +2, and + 3 cm H2O. Each pressure support level was maintained for 20 min. Esophageal pressure was monitored. Pressure-time products of respiratory muscle and ventilator insufflation were measured, and the fraction of pressure generated by the patient was calculated to represent the contribution of the patient's inspiratory effort. A total of 105 datasets were collected at different PMI-targeted pressure support levels. The differences in PMI between the target and the obtained value were all within ±1 cm H2O. As targeted PMI increased, pressure support settings decreased significantly from a median (interquartile range) of 11 (10-12) to 5 (4-6) cm H2O (p < 0.001), which resulted in a significant increase in pressure-time products of respiratory muscle [from 2.9 (2.1-5.0) to 6.8 (5.3-8.1) cm H2O•s] and the fraction of pressure generated by the patient [from 25% (19-31%) to 72% (62-87%)] (p < 0.001). The area under receiver operating characteristic curves for PMI to predict 30 and 70% contribution of patient's effort were 0.93 and 0.95, respectively. High sensitivity (all 1.00), specificity (0.86 and 0.78), and negative predictive value (all 1.00), but low positive predictive value (0.61 and 0.43) were obtained to predict either high or low contribution of patient's effort. Our results preliminarily suggested the feasibility of pressure support adjustment according to the PMI target from the ventilator screen. PMI could reliably predict the high and low contribution of a patient's effort during assisted ventilation.Clinical trial registration: ClinicalTrials.gov, identifier NCT05970393.

Prevalence of Dysbetalipoproteinemia in the UK Biobank According to Different Diagnostic Criteria

Abstract Context Dysbetalipoproteinemia (DBL) is a multifactorial disorder that disrupts the normal metabolism of remnant lipoproteins, causing increased risk of cardiovascular disease. However, establishing a proper diagnosis is difficult, and the true prevalence of the disease in the general population remains unknown. Objective The objectives were to study the prevalence of the disease and to validate the performance of different clinical diagnostic criteria in a large population-based cohort. Methods This study included 453 437 participants from the UK Biobank. DBL was established in participants having an ε2ε2 genotype with mixed dyslipidemia or lipid-lowering therapy use (n = 964). The different diagnostic criteria for DBL were applied in individuals without lipid-lowering medication (n = 370 039, n = 534 DBL), to compare their performance. Results Overall, 0.6% of participants had an ε2ε2 genotype, of which 36% were classified as DBL, for a disease prevalence of 0.2% (1:469). The prevalence of DBL was similar between the different genetic ancestries (≤0.2%). Several diagnostic criteria showed good sensitivity for the diagnosis of DBL (&amp;gt;90%), but they suffered from a very low positive predictive value (0.6-15.4%). Conclusion This study reported for the first time the prevalence of DBL in the UK Biobank according to genetic ancestry. Furthermore, we provided the first external validation of different diagnostic criteria for DBL in a large population-based cohort and highlighted the fact that these criteria should not be used to diagnose DBL alone but should rather be used as a first screening step to determine which individuals may benefit from genetic testing to confirm the diagnosis.

A Secondary Retrospective Analysis of the Predictive Value of Neutrophil-Reactive Intensity (NEUT-RI) in Septic and Non-Septic Patients in Intensive Care.

Effective identification and management in the early stages of sepsis are critical for achieving positive outcomes. In this context, neutrophil-reactive intensity (NEUT-RI) emerges as a promising and easily interpretable parameter. This study aimed to assess the predictive value of NEUT-RI in diagnosing sepsis and to evaluate its prognostic significance in distinguishing 28-day mortality outcomes. This study is a secondary, retrospective, observational analysis. Clinical data upon ICU admission were collected. We enrolled septic patients and a control group of critically ill patients without sepsis criteria. The patients were divided into subgroups based on renal function for biomarker evaluation with 28-day outcomes reported for septic and non-septic patients. A total of 200 patients were included in this study. A significant difference between the "septic" and "non-septic" groups was detected in the NEUT-RI plasma concentration (53.80 [49.65-59.05] vs. 48.00 [46.00-49.90] FI, p < 0.001, respectively). NEUT-RI and procalcitonin (PCT) distinguished between not complicated sepsis and septic shock (PCT 1.71 [0.42-12.09] vs. 32.59 [8.83-100.00], <0.001 and NEUT-RI 51.50 [47.80-56.30] vs. 56.20 [52.30-61.92], p = 0.005). NEUT-RI, PCT, and CRP values were significantly different in patients with "renal failure". NEUT-RI and PCT at admission in the ICU in the septic group were higher in patients who died (58.80 [53.85-73.10] vs. 53.05 [48.90-57.22], p = 0.005 and 39.56 [17.39-83.72] vs. 3.22 [0.59-32.32], p = 0.002, respectively). Both NEUT-RI and PCT showed a high negative predictive value and low positive predictive value. The inflammatory biomarkers assessed in this study offer valuable support in the early diagnosis of sepsis and could have a possible role in anticipating the outcome. NEUT-RI elevation appears particularly promising for early sepsis detection and severity discrimination upon admission.

Large language models facilitate the generation of electronic health record phenotyping algorithms.

Phenotyping is a core task in observational health research utilizing electronic health records (EHRs). Developing an accurate algorithm demands substantial input from domain experts, involving extensive literature review and evidence synthesis. This burdensome process limits scalability and delays knowledge discovery. We investigate the potential for leveraging large language models (LLMs) to enhance the efficiency of EHR phenotyping by generating high-quality algorithm drafts. We prompted four LLMs-GPT-4 and GPT-3.5 of ChatGPT, Claude 2, and Bard-in October 2023, asking them to generate executable phenotyping algorithms in the form of SQL queries adhering to a common data model (CDM) for three phenotypes (ie, type 2 diabetes mellitus, dementia, and hypothyroidism). Three phenotyping experts evaluated the returned algorithms across several critical metrics. We further implemented the top-rated algorithms and compared them against clinician-validated phenotyping algorithms from the Electronic Medical Records and Genomics (eMERGE) network. GPT-4 and GPT-3.5 exhibited significantly higher overall expert evaluation scores in instruction following, algorithmic logic, and SQL executability, when compared to Claude 2 and Bard. Although GPT-4 and GPT-3.5 effectively identified relevant clinical concepts, they exhibited immature capability in organizing phenotyping criteria with the proper logic, leading to phenotyping algorithms that were either excessively restrictive (with low recall) or overly broad (with low positive predictive values). GPT versions 3.5 and 4 are capable of drafting phenotyping algorithms by identifying relevant clinical criteria aligned with a CDM. However, expertise in informatics and clinical experience is still required to assess and further refine generated algorithms.

Manual method versus flow cytometry for diagnosing spontaneous bacterial peritonitis

ABSTRACT Background Diagnosing spontaneous bacterial peritonitis (SBP) requires a high clinical index of suspicion because the clinical presentation varies widely. Early detection and treatment of SBP are crucial for improving survival rates. Ascitic fluid neutrophil count ≥ 250 cells/mm3 is the only required criterion to diagnose SBP, regardless of culture results. The most frequently used manual ascitic neutrophil counting is difficult and time-consuming, and it often leads to frequent delays in obtaining results and a possible under- or overestimation of the neutrophilic count. The flow cytometric counting provides results in 30 minutes and is more accurate, with sensitivity and specificity approaching 100%. Objective This study aimed to assess the diagnostic efficacy of manual counting of neutrophils in comparison to flow cytometry for detecting SBP. Patients and methods In a cross-sectional study, we included 140 cirrhosis-related ascites patients who were clinically suspected to have SBP and admitted to the Alexandria Main University Hospital in the Department of Internal Medicine. Participants with ≥ 250 neutrophil cells/mm3 in ascitic fluid by flow cytometry were defined as SBP patients. In contrast, those with fewer than 250 neutrophil cells/mm3 in ascitic fluid using the flow cytometric method were defined as patients without SBP. Results At a cutoff value of 250 cells/mm3, the ROC curve showed that the manual ascitic fluid method had lower sensitivity, specificity, positive and negative predictive values (67.19%, 86.84%, 81.13%, and 75.86%, respectively, with AUC = 0.945, 95% CI, 0.91–0.97, p < 0.001) in diagnosing SBP than the flow cytometric method. Conclusions Using the flow cytometric method significantly improved diagnostic precision. Its use may be recommended to replace the traditional manual method, which may have unaccepted false-negative results.

Abnormal biochemical indicators of neonatal inherited metabolic disease in carriers

BackgroundTraditional biochemical screening for neonatal inherited metabolic diseases has high false-positive rates and low positive predictive values, which are not conducive to early diagnosis and increase parents’ anxiety. This study analysed the relationship between gene variant carriers and their biochemical indicators in traditional biochemical screening, aiming to find explanations for false positives in newborns.ResultsThis retrospective study included 962 newborns. Newborns underwent traditional biochemical screening at birth using blood staining and genomic sequencing of their stored blood staining using the NeoSeq Pro panel, which was able to detect 154 pathogenic genes and 86 diseases. A total of 632 newborns were carriers of gene variants. 56% of congenital hypothyroidism carriers had higher thyroid-stimulating hormone levels than normal newborns. Abnormal biochemical indices were detected in 71% of carriers of organic acid metabolic diseases, 69% of carriers of amino acid metabolic diseases, and 85% of carriers of fatty acid β oxidation disorders. In carriers associated with organic acid metabolic diseases, the propionylcarnitine (C3), C3/acetylcarnitine (C2), and methylmalonylcarnitine (C4DC) + 3-hydroxyisovalerylcarnitine (C5OH) levels were higher than those in non-carriers (C3: 4.12 vs. 1.66 µmol/L; C3/C2: 0.15 vs. 0.09; C4DC + C5OH: 0.22 vs. 0.19 µmol/L). In carriers associated with amino acid metabolic diseases, phenylalanine levels were higher than those in non-carriers (68.00 vs. 52.05 µmol/L). For carriers of fatty acid β oxidation disorders, butyrylcarnitine levels were higher than those in non-carriers (0.31 vs. 0.21 µmol/L), while the free carnitine levels were lower than those in non-carriers (14.65 vs. 21.87 µmol/L). There was a higher occurrence of carriers among newborns who received false-positive results for amino acid metabolic diseases compared to those who received negative results (15.52% vs. 6.71%). Similarly, there was a higher occurrence of carriers among newborns who received false-positive results for fatty acid β oxidation disorders compared to those who received negative results (28.30% vs. 7.29%).ConclusionsThis study showed that the carriers comprised a large number of newborns. Carriers had abnormal biochemical indicators compared with non-carriers, which could explain the false-positive rate for newborns using traditional newborn biochemical screening, especially in amino acid metabolic and fatty acid β oxidation disorders.

Non-invasive assessment of left ventricular filling pressure in aortic stenosis.

The assessment of left ventricular (LV) filling pressure (FP) is important for the management of aortic stenosis (AS) patients. Although, it is often restricted for predict LV FP in AS because of mitral annular calcification and a certain left ventricular hypertrophy. Thus, we tested the predictive ability of the algorithm for elevated LV FP in AS patients and also applied a recently-proposed echocardiographic scoring system of LV FP, visually assessed time difference between the mitral valve and tricuspid valve opening (VMT) score. We enrolled consecutive 116 patients with at least moderate AS in sinus rhythm who underwent right heart catheterization and echocardiography within 7 days. Mean pulmonary artery wedge pressure (PAWP) was measured as invasive parameter of LV FP. LV diastolic dysfunction (DD) was graded according to the ASE/EACVI guidelines. The VMT score was defined as follows: time sequence of opening of mitral and tricuspid valves was scored to 0-2 (0: tricuspid valve first, 1: simultaneous, 2: mitral valve first). When the inferior vena cava was dilated, one point was added and VMT score was finally calculated as 0-3. Of the 116 patients, 29 patients showed elevated PAWP. Ninety patients (93%) and 67 patients (63%) showed increased values for left atrium volume index (LAVI) and E/e', respectively when the cut-off values recommended by the guidelines were applied and thus the algorism predicted elevated PAWP with a low specificity and positive predictive value (PPV). VMT≥2 predicted elevated PAWP with a sensitivity of 59%, specificity of 90%, PPV of 59%, and negative predictive value of 89%. An alternative algorithm that applied tricuspid regurgitation velocity and VMT scores was tested, and its predictive ability was markedly improved. VMT score was applicable for AS patients. Alternative use of VMT score improved diagnostic accuracy of guideline-recommended algorism.

Remote screening accuracy of first-time hearing aid users.

Capacity constraints in Danish hearing healthcare may lead to diagnostic delays and repetitive pre-treatment audiological exams for hearing-impaired patients. This study investigated the effectiveness of remote ear-nose-throat (ENT) specialist assessments (RESA) for complicated hearing loss, comparing the accuracy of private ENT specialists and medical audiologists. RESA screening accuracy was determined for four ENT specialists, individually and as subspecialised groups. These assessments were benchmarked against "gold standard" in-person ENT assessments for 445 potential adult first-time hearing aid users. Medical audiologists initially recorded lower RESA screening specificity and positive predictive values than private ENT specialists. However, after making two adjustments to the dataset, these differences were neutralised. Screening sensitivity was consistent across individual and grouped subspecialities. RESA screening is a promising tool for timely diagnosis and treatment. The findings reveal that both private ENT specialists and medical audiologists may conduct RESA with high consistency and uniformity. This research was funded by the Danish Health Data Authority and the North Denmark Region, but does not reflect their views. The study received no commercial support. Not relevant.

Prediction of Short-Term Postoperative Complications Following Open Reduction Internal Fixation of Acute Ankle Fractures

Introduction/Purpose: Acute ankle fractures are a common injury and represent a significant financial burden to the healthcare system. Reimbursement following ankle fractures is complicated by bundled payments within the 90-day global period which do not account for unpredictable complications. Comorbidity indices, such as the American Society of Anesthesiologists (ASA) classification, are often correlated with complications; however, the positive predictive value of these metrics have not been investigated. As insurers continue to transition towards bundled reimbursements, greater attentiveness towards providing efficient care is paramount. The purpose of this investigation is to develop a machine learning (ML) model that is predictive of complications following operative management of acute ankle fractures. This study will also compare ML algorithms to legacy indices to assess the predictive value in predicting complications. Methods: The American College of Surgeons-National Surgical Quality Improvement Program database was queried via Current Procedural Terminology (CPT) from 2015-2018 for open reduction internal fixation (ORIF) cases for ankle fractures, including ORIF of the medial malleolus, posterior malleolus, lateral malleolus, bimalleolar, trimalleolar, syndesmosis. Training and validation sets were created by randomly assigning 80% and 20% of the data set. Included variables were age, body mass index (BMI), operative time, smoking status, comorbidities, diagnosis, and preoperative hematocrit and albumin. Complications included any adverse event, transfusion, surgical site infection, return to the operating room, deep vein thrombosis or pulmonary embolism, pneumonia, urinary tract infection, cerebrovascular accident, cardiac arrest, myocardial infarction. Each ML algorithm was compared with one another and to a baseline model using American Society of Anesthesiologists (ASA) classification. Model strength was evaluated by calculating the area under the curve (AUC) for receiver operating characteristic (ROC) of any adverse event. Results: We identified a total of 28,736 ORIF cases. Mean age and BMI were 32.2 ± 18.6 years, 30.8 ± 7.6 kg/m^2, respectively. The overall incidence of any medical complication was 3.4% (n=975) with surgical site infection (1.3%, n=386) and return to operating room (1.3%, n=367) being the most common. Percentage hematocrit, BMI, operative time, and age were of highest importance in outcome prediction. Logistic regression (AUC: 72%) and gradient boosting (AUC: 72%) ML algorithms outperformed ASA classification (AUC: 69%) for predicting any adverse event following ORIF of ankle fractures. Conclusion: Legacy comorbidity indices are simple metrics that can be easily constructed from patient demographic variables and past medical history. However, the predictive power is limited as they have shown a low positive predictive value for predicting complications. Machine learning algorithms can calculate patient-specific risk for postoperative complications which may adjust perioperative care and site of surgery. These models have the capability to optimize utilization of healthcare resources and minimize excessive expenditure by providing patient-specific care. Receiver Operating Curve-Area under the Curve Analysis for Adverse Events Following ORIF of Ankle Fractures

Predicting the severity of acute pancreatitis: Current approaches and future directions

Acute pancreatitis (AP) is a sudden-onset inflammatory disease of the pancreas. The severity of AP is classified into mild, moderate, and severe categories based on the presence and persistence of organ failure. Severe acute pancreatitis (SAP) can be associated with significant morbidity and mortality. It requires early recognition for appropriate timely management. Prognostic scores for predicting SAP incorporating many clinical, laboratory, and radiological parameters have been developed in the past. However, all of these prognostic scores have low positive predictive value for SAP and some of these scores require >24 h for assessment. There is a need to develop biomarkers that can accurately identify patients at risk for SAP early in the course of the presentation. In this review, we aim to provide a summary of the most commonly utilized prognostic scores for AP and discuss future directions.