Machine Learning for Predictive Analysis of Otolaryngology Residency Letters of Recommendation.

INTRODUCTION: Letters of recommendation (LORs) are a highly influential yet subjective and often enigmatic aspect of the residency application process. This study hypothesizes that LORs do contain valuable insights into applicants and can be used to predict outcomes. This pilot study utilizes natural language processing and machine learning (ML) models using LOR text to predict interview invitations for otolaryngology residency applicants. METHODS: A total of 1642 LORs from the 2022-2023 application cycle were retrospectively retrieved from a single institution. LORs were preprocessed and vectorized using three different techniques to represent the text in a way that an ML model can understand written prose: CountVectorizer (CV), Term Frequency-Inverse Document Frequency (TF-IDF), and Word2Vec (WV). Then, the LORs were trained and tested on five ML models: Logistic Regression (LR), Naive Bayes (NB), Decision Tree (DT), Random Forest (RF), and Support Vector Machine (SVM). RESULTS: Of the 337 applicants, 67 were interviewed and 270 were not interviewed. In total, 1642 LORs (26.7% interviewed) were analyzed. The two best-performing ML models in predicting interview invitations were the TF-IDF vectorized DT and CV vectorized DT models. CONCLUSION: This preliminary study revealed that ML models and vectorization combinations can provide better-than-chance predictions for interview invitations for otolaryngology residency applicants. The high-performing ML models were able to classify meaningful information from the LORs to predict applicant interview invitation. The potential of an automated process to help predict an applicant's likelihood of obtaining an interview invitation could be a valuable tool for training programs in the future. LEVEL OF EVIDENCE: N/A Laryngoscope, 2024.

Characterization of Listeria Monocytogenes Originating from the Spanish Meat-Processing Chain.

Using agglutination techniques, 118 Listeria monocytogenes isolates from red meat and poultry were serotyped. Strains were ascribed to the serotypes 4b/4e (44.1% of the strains), 1/2 (a, b or c; 28.0%), 4c (6.8%), 4d/4e (5.9%) and 3 (a, b or c; 2.5%). Among these are the serotypes most frequently involved in cases of human listeriosis. The susceptibility of 72 strains to 26 antibiotics of clinical importance was determined by disc diffusion (Clinical and Laboratory Standards Institute; CLSI). High levels of resistance were observed to cefoxitin (77.8% of the strains showed resistance), cefotaxime (62.5%), cefepime (73.6%), and nalidixic acid (97.2%), nitrofurantoin (51.4%) and oxacillin (93.1%). Less than 3% of the strains showed resistance to the antibiotic classes used in human listeriosis therapy (i.e., ampicillin, gentamicin, rifampicin, chloramphenicol, enrofloxacin, vancomycin, trimethoprim-sulfamethoxazole, erythromycin, and tetracycline). The influence of species and serotype on the growth kinetics (modified Gompertz equation) and on the adhesion ability (crystal violet staining) of nine isolates of L. monocytogenes (serotypes 1/2a, 1/2b, 1/2c, 3a, 3b, 3c, 4a, 4b, and 4d), and one strain of Listeria ivanovii were investigated. The maximum growth rate (ΔOD420-580/h) varied between 0.073 ± 0.018 (L. monocytogenes 1/2a) and 0.396 ± 0.026 (L. monocytogenes 4b). The isolates of L. monocytogenes belonging to serotypes 3a and 4a, as well as L. ivanovii, showed a greater (p < 0.05) biofilm-forming ability than did the remaining strains, including those that belong to the serotypes commonly implied in human listeriosis (1/2a, 1/2b, 1/2c and 4b). The need for training in good hygiene practices during the handling of meat and poultry is highlighted to reduce the risk of human listeriosis.

What are the associations between the quantity of faculty evaluations and residents' perception of quality feedback?

OBJECTIVES: To determine if there is a correlation between the numbers of evaluations submitted by faculty and the perception of the quality of feedback reported by trainees on a yearly survey. METHOD: 147 ACGME-accredited training programs sponsored by a single medical school were included in the analysis. Eighty-seven programs (49 core residency programs and 38 advanced training programs) with 4 or more trainees received ACGME survey summary data for academic year 2013-2014. Resident ratings of satisfaction with feedback were analyzed against the number of evaluations completed per resident during the same period. R-squared correlation analysis was calculated using a Pearson correlation coefficient. RESULTS: 177,096 evaluations were distributed to the 87 programs, of which 117,452 were completed (66%). On average, faculty submitted 33.9 evaluations per resident. Core residency programs had a greater number of evaluations per resident than fellowship programs (39.2 vs. 27.1, respectively, p = 0.15). The average score for the "satisfied with feedback after assignment" survey questions was 4.2 (range 2.2-5.0). There was no overall correlation between the number of evaluations per resident and the residents' perception of feedback from faculty based on medical, surgical or hospital-based programs. CONCLUSIONS: Resident perception of feedback is not correlated with number of faculty evaluations. An emphasis on faculty summative evaluation of resident performance is important but appears to miss the mark as a replacement for on-going, data-driven, structured resident feedback. Understanding the difference between evaluation and feedback is a global concept that is important for all medical educators and learners.