DukeSpace

Find and submit publications, theses, and dissertations by Duke faculty and students.

 

Communities in DukeSpace

Select a community to browse its collections.

Recent Submissions

ItemOpen Access
Incorporating informatively collected laboratory data from EHR in clinical prediction models.
(BMC medical informatics and decision making, 2024-07) Sun, Minghui; Engelhard, Matthew M; Bedoya, Armando D; Goldstein, Benjamin A

Background

Electronic Health Records (EHR) are widely used to develop clinical prediction models (CPMs). However, one of the challenges is that there is often a degree of informative missing data. For example, laboratory measures are typically taken when a clinician is concerned that there is a need. When data are the so-called Not Missing at Random (NMAR), analytic strategies based on other missingness mechanisms are inappropriate. In this work, we seek to compare the impact of different strategies for handling missing data on CPMs performance.

Methods

We considered a predictive model for rapid inpatient deterioration as an exemplar implementation. This model incorporated twelve laboratory measures with varying levels of missingness. Five labs had missingness rate levels around 50%, and the other seven had missingness levels around 90%. We included them based on the belief that their missingness status can be highly informational for the prediction. In our study, we explicitly compared the various missing data strategies: mean imputation, normal-value imputation, conditional imputation, categorical encoding, and missingness embeddings. Some of these were also combined with the last observation carried forward (LOCF). We implemented logistic LASSO regression, multilayer perceptron (MLP), and long short-term memory (LSTM) models as the downstream classifiers. We compared the AUROC of testing data and used bootstrapping to construct 95% confidence intervals.

Results

We had 105,198 inpatient encounters, with 4.7% having experienced the deterioration outcome of interest. LSTM models generally outperformed other cross-sectional models, where embedding approaches and categorical encoding yielded the best results. For the cross-sectional models, normal-value imputation with LOCF generated the best results.

Conclusion

Strategies that accounted for the possibility of NMAR missing data yielded better model performance than those did not. The embedding method had an advantage as it did not require prior clinical knowledge. Using LOCF could enhance the performance of cross-sectional models but have countereffects in LSTM models.
ItemOpen Access
Hematology and serum biochemistry of free-ranging mantled howler monkeys (Alouatta palliata) at La Pacifica, Costa Rica.
(Journal of medical primatology, 2024-08) Larsen, R Scott; Moresco, Anneke; Meneses, Ana; Glander, Kenneth E

Background

Hematologic and blood biochemical values are key tools for assessing primate health. A long-term behavioral study of howler monkeys at a single site (La Pacífica, Guanacaste, Costa Rica), afforded the opportunity to develop baseline values for a large group of animals, evaluating differences between adult males and females and comparing to a report in the same population two decades later.

Methods

In 1998, 64 free-ranging mantled howler monkeys were anesthetized and sampled for hematologic and biochemical analysis.

Results

Blood analysis is reported for 29 adult females, 9 juvenile females, 19 adult males and 3 juvenile males. Four adults were excluded due to external injury or disease. There were few significant differences between adult females, juvenile females, and adult males.

Conclusions

Baseline blood parameters are useful for determining normal values for howler monkey populations. The values for total protein, blood urea nitrogen, glucose, liver enzymes and potassium differed from a later study in 2019 may indicate changes that are influencing howler monkey health.
ItemOpen Access
Serosurvey of Arboviruses in Free-ranging Mantled Howler Monkeys (Alouatta palliata) in Costa Rica.
(Journal of wildlife diseases, 2024-11) Larsen, R Scott; Moresco, Anneke; Karabatsos, Nick; Dolz, Gaby; Glander, Kenneth E
We investigated the prevalence of arthropod-borne viral diseases in a population of free-ranging mantled howler monkeys (Alouatta palliata) in Costa Rica in 1998. Blood samples were opportunistically collected from monkeys anesthetized for another study. Serology was performed on 64 individuals to assess exposure of this population to vesicular stomatitis virus, equine encephalitis viruses, Mayaro virus, St. Louis encephalitis virus, yellow fever virus, and dengue virus. The New Jersey serotype of vesicular stomatitis (VSV-NJ) was the only pathogen for which the population tested positive (44% [28/64]). This is the first report of antibodies against VSV-NJ in nonhuman primates in Costa Rica.
ItemOpen Access
New Observations on the Restoration of Notre-Dame in Paris
(Journal of the Society of Architectural Historians, 2023-12-01) Bruzelius, C
ItemOpen Access
:Dominicans and Franciscans in Medieval Rome: History, Architecture, and Art
(Speculum, 2024-10-01) Bruzelius, Caroline