Browsing by Subject "prediction"
Now showing 1 - 8 of 8
Results Per Page
Sort Options
Item Aggregate Financial Misreporting and the Predictability of U.S. Recessions(SSRN, 2021) Beneish, Messod D.; Farber, David B.; Glendening, Matthew; Shaw, Kenneth W.; Kelley School of Business - IndianapolisWe rely on the theoretical prediction that financial misreporting peaks before economic busts to examine whether aggregate ex ante measures of the likelihood of financial misreporting improve the predictability of U.S. recessions. We consider six measures of misreporting and show that the Beneish M-Score significantly improves out-of-sample recession prediction at longer forecasting horizons. Specifically, relative to leading models based on yield spreads and market returns, M-Score increases the average probability of a recession across forecast horizons of six-, seven-, and eight-quarters-ahead by 56 percent, 79 percent, and 92 percent, respectively. These findings are robust to alternative definitions of interest rate spreads, and to controlling for the federal funds rate, investor sentiment, and aggregate earnings growth. We show that the performance of M-Score likely arises because firms with high M-Scores tend to experience negative future performance. Overall, this study provides novel evidence that accounting information can be useful to forecasters and regulators interested in assessing the likelihood of U.S. recessions a few quarters ahead.Item Aggregate Financial Misreporting and the Predictability of U.S. Recessions and GDP Growth(American Accounting Association, 2023-09-01) Beneish, Messod D.; Farber, David B.; Glendening, Matthew; Shaw, Kenneth W.; Kelley School of BusinessThis study examines the incremental predictive power of aggregate measures of financial misreporting for recession and real gross domestic product (GDP) growth. We draw on prior research suggesting that misreporting has real economic effects because it represents misinformation on which firms base their investment, hiring, and production decisions. We find that aggregate M-Score incrementally predicts recessions at forecast horizons of five to eight quarters ahead. We also find that aggregate M-Score is significantly associated with lower future growth in real GDP, real investment, consumption, and industrial production. Additionally, our result that aggregate M-Score predicts lower real investment one to four quarters ahead partially accounts for why misreporting predicts recessions five to eight quarters ahead. Our findings are weaker when we use aggregate F-Score as a proxy for misreporting. Overall, this study provides novel evidence that aggregate misreporting measures can aid forecasters and regulators in predicting recessions and real GDP growth.Item Are there specific antepartum factors and labor complications that predict elevated immediate postpartum Edinburgh Postpartum Depression Scale scores?(2019-10-09) Ayo, Katherine V; Teal, Evgenia; Haas, David MBackground: Postpartum depression is a common medical condition diagnosed in the weeks after delivery. There are several modifiable risk factors during the antepartum, labor and delivery, and immediate postpartum periods that may influence the likelihood that an individual will develop this condition. Objective: The purpose of this study was to identify risk factors that may predict an individual’s risk of developing postpartum depression. Methods: We conducted a retrospective cohort study of all deliveries over a 14 month period. Demographic characteristics and complications during pregnancy and delivery were obtained from the electronic medical record. The Edinburg Perinatal Depression Scale (EPDS) was administered to all postpartum women before discharge. Antenatal and delivery characteristic associations with EPDS cutoffs of ≥10 and ≥13 were determined and significant variables were included in a logistic regression to determine predictive factors for elevated immediate postpartum EPDS scores. Results: A total of 1,913 women had valid immediate postpartum EPDS results. Women with a history of depression, those with a positive drug screen on admission to labor and delivery, those with babies admitted to the neonatal intensive care unit (NICU), and those with alcohol or opioid abuse were found to have increased risk of development of PPD. Logistic regression analysis found that having a positive drug screen (OR 2.54, 95% CI 1.43-4.52) history of depression (OR 3.97, 95% CI 2.44-6.30), alcohol use (OR 5.30, 95% CI 1.39-20.16), and opioid use disorder (OR 8.64, 95% CI 1.06-70.49) predicted EPDS scores ≥10, while having a baby admitted to the NICU (OR 1.70, 95% CI 1.20-2.57), history of depression (OR 4.46, 95% CI 2.81-7.07), opioid use disorder (OR 9.32, 95% CI 1.14-76.39) predicted EPDS scores ≥13. Conclusion: Several modifiable risk factors were found that could lead to an increased risk of PPD. Early screening and intervention based on risk factors may decrease the likelihood of developing early postpartum depression.Item Clinical features from the history and physical examination that predict the presence or absence of pulmonary embolism in symptomatic emergency department patients: results of a prospective, multi-center study(2010-04) Courtney, Mark; Kline, Jeffrey A.; Kabrhel, Christopher; Moore, Christopher L; Smithline, Howard A; Nordenholz, Kristen E; Richman, Peter B; Plewa, Michael CStudy Objective—Prediction rules for pulmonary embolism (PE) employ variables explicitly shown to estimate the probability of PE. However, clinicians often use variables that have not been similarly validated, yet are implicitly believed to modify probability of PE. The objective of this study was to measure the predictive value of 13 implicit variables. Methods—Patients were enrolled in a prospective cohort study from 12 centers in the United States; all had an objective test for PE (D-dimer, CT angiography, or V/Q scan). Clinical features including 12 predefined previously validated (explicit) variables and 13 variables not part of existing prediction rules (implicit) were prospectively recorded at presentation. The primary outcome was VTE (venous thromboembolism: PE or deep venous thrombosis), diagnosed by imaging up to 45 days after enrollment. Variables with adjusted odds ratios from logistic regression with 95% confidence intervals not crossing unity were considered significant. Results—7,940 patients (7.2% VTE+) were enrolled. Mean age was 49±17 years and 67% were female. Eight of 13 implicit variables were significantly associated with VTE; those with an adjusted OR >1.5 included non-cancer related thrombophilia (1.99), pleuritic chest pain (1.53), and family history of VTE (1.51). Implicit variables that predicted no VTE outcome included: substernal chest pain, female gender, and smoking. Nine of 12 explicit variables predicted a positive outcome of VTE, including unilateral leg swelling, recent surgery, estrogen, hypoxemia and active malignancy. Conclusions—In symptomatic outpatients being considered for possible PE, non-cancer related thrombophilia, pleuritic chest pain, and family history of VTE increase probability of PE or DVT. Other variables that are part of existing pretest probability systems were validated as important predictors in this diverse sample of US Emergency department patients.Item Joint models for longitudinal and survival data(2014-07-11) Yang, Lili; Gao, Sujuan; Yu, Menggang; Tu, Wanzhu; Callahan, Christopher M.; Zollinger, TerrellEpidemiologic and clinical studies routinely collect longitudinal measures of multiple outcomes. These longitudinal outcomes can be used to establish the temporal order of relevant biological processes and their association with the onset of clinical symptoms. In the first part of this thesis, we proposed to use bivariate change point models for two longitudinal outcomes with a focus on estimating the correlation between the two change points. We adopted a Bayesian approach for parameter estimation and inference. In the second part, we considered the situation when time-to-event outcome is also collected along with multiple longitudinal biomarkers measured until the occurrence of the event or censoring. Joint models for longitudinal and time-to-event data can be used to estimate the association between the characteristics of the longitudinal measures over time and survival time. We developed a maximum-likelihood method to joint model multiple longitudinal biomarkers and a time-to-event outcome. In addition, we focused on predicting conditional survival probabilities and evaluating the predictive accuracy of multiple longitudinal biomarkers in the joint modeling framework. We assessed the performance of the proposed methods in simulation studies and applied the new methods to data sets from two cohort studies.Item Predicting Dementia With Routine Care EMR Data(Elsevier, 2020-01) Ben Miled, Zina; Haas, Kyle; Black, Christopher M.; Khandker, Rezaul Karim; Chandrasekaran, Vasu; Lipton, Richard; Boustani, Malaz A.; Electrical and Computer Engineering, School of Engineering and TechnologyOur aim is to develop a machine learning (ML) model that can predict dementia in a general patient population from multiple health care institutions one year and three years prior to the onset of the disease without any additional monitoring or screening. The purpose of the model is to automate the cost-effective, non-invasive, digital pre-screening of patients at risk for dementia. Towards this purpose, routine care data, which is widely available through Electronic Medical Record (EMR) systems is used as a data source. These data embody a rich knowledge and make related medical applications easy to deploy at scale in a cost-effective manner. Specifically, the model is trained by using structured and unstructured data from three EMR data sets: diagnosis, prescriptions, and medical notes. Each of these three data sets is used to construct an individual model along with a combined model which is derived by using all three data sets. Human-interpretable data processing and ML techniques are selected in order to facilitate adoption of the proposed model by health care providers from multiple institutions. The results show that the combined model is generalizable across multiple institutions and is able to predict dementia within one year of its onset with an accuracy of nearly 80% despite the fact that it was trained using routine care data. Moreover, the analysis of the models identified important predictors for dementia. Some of these predictors (e.g., age and hypertensive disorders) are already confirmed by the literature while others, especially the ones derived from the unstructured medical notes, require further clinical analysis.Item Quality Assurance of Computer-Aided Detection and Diagnosis in Colonoscopy(Elsevier, 2019) Vinsard, Daniela Guerrero; Mori, Yuichi; Misawa, Masashi; Kudo, Shin-ei; Rastogi, Amit; Bagci, Ulas; Rex, Douglas K.; Wallace, Michael B.; Medicine, School of MedicineRecent breakthroughs in artificial intelligence (AI), specifically via its emerging sub-field “Deep Learning,” have direct implications for computer-aided detection and diagnosis (CADe/CADx) for colonoscopy. AI is expected to have at least 2 major roles in colonoscopy practice; polyp detection (CADe) and polyp characterization (CADx). CADe has the potential to decrease polyp miss rate, contributing to improving adenoma detection, whereas CADx can improve the accuracy of colorectal polyp optical diagnosis, leading to reduction of unnecessary polypectomy of non-neoplastic lesions, potential implementation of a resect and discard paradigm, and proper application of advanced resection techniques. A growing number of medical-engineering researchers are developing both, CADe and CADx systems, some of which allow real-time recognition of polyps or in vivo identification of adenomas with over 90% accuracy. However, the quality of the developed AI systems as well as that of the study designs vary significantly, hence raising some concerns regarding the generalization of the proposed AI systems. Initial studies were conducted in an exploratory or retrospective fashion using stored images and likely overestimating the results. These drawbacks potentially hinder smooth implementation of this novel technology into colonoscopy practice. The aim of this article is to review both contributions and limitations in recent machine learning based CADe/CADx colonoscopy studies and propose some principles that should underlie system development and clinical testing.Item Time to Peak Glucose and Peak C-Peptide During the Progression to Type 1 Diabetes in the Diabetes Prevention Trial and TrialNet Cohorts(ADA, 2021-10) Voss, Michael G.; Cleves, Mario M.; Cuthbertson, David D.; Xu, Ping; Evans-Molina, Carmella; Palmer, Jerry P.; Redondo, Maria J.; Steck, Andrea K.; Lundgren, Markus; Larsson, Helena; Moore, Wayne V.; Atkinson, Mark A.; Sosenko, Jay; Ismail, Heba M.; Pediatrics, School of MedicineObjective: To assess the progression of type 1 diabetes using time to peak glucose or C-peptide during oral glucose tolerance tests (OGTTs) in autoantibody positive (Ab+) relatives of people with type 1 diabetes. Methods: We examined 2-hour OGTTs of participants in the Diabetes Prevention Trial Type 1 (DPT-1) and TrialNet Pathway to Prevention (PTP) studies. We included 706 DPT-1 participants (Mean±SD age: 13.84±9.53 years; BMI-Z-Score: 0.33±1.07; 56.1% male) and 3,720 PTP participants (age: 16.01±12.33 Years, BMI-Z-Score 0.66±1.3; 49.7% male). Log-rank testing and Cox regression analyses with adjustments (age, sex, race, BMI-Z-Score and peak Glucose/Cpeptide levels, respectively) were performed. Results: In each of DPT-1 and PTP, higher 5-year risk of diabetes development was seen in those with time to peak glucose >30 min and time to peak C-peptide >60 min (p<0.001 for all groups), before and after adjustments. In models examining strength of association with diabetes development, associations were greater for time to peak C-peptide versus peak C-peptide value (DPT-1: X2 = 25.76 vs. X2 = 8.62 and PTP: X2 = 149.19 vs. X2 = 79.98; all p<0.001). Changes in the percentage of individuals with delayed glucose and/or C-peptide peaks were noted over time. Conclusions: In two independent at risk populations, we show that those with delayed OGTT peak times for glucose or C-peptide are at higher risk of diabetes development within 5 years, independent of peak levels. Moreover, time to peak C-peptide appears more predictive than the peak level, suggesting its potential use as a specific biomarker for diabetes progression.