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, 134:4016-4022, 2024.

Evaluating human papillomavirus testing, prevalence, and association with prognosis in head and neck squamous cell carcinoma by subsite: A national cancer database study.

PURPOSE: To compare human papillomavirus (HPV) testing, prevalence, and association with prognosis between head and neck squamous cell carcinoma (HNSCC) subsites. MATERIALS AND METHODS: This study utilized the National Cancer Database (NCDB) to identify patients diagnosed with HNSCC between 2010 and 2017. Rates of HPV testing, HPV-positivity, and changes in these rates over time were measured by subsite. The impact of HPV-positivity on overall survival across six head and neck subsites was assessed using multivariable-adjusted Cox proportional hazards analysis. RESULTS: A total of 121,550 patients were included. Of this cohort, 87,575 (72.1%) were tested for HPV, with the oropharynx (55,049/64,158; 85.8%) displaying the highest rates of testing and the sinonasal tract (1519/2853; 53.2%) displaying the lowest testing rates. Of the 86,136 with a definitive result, 46,878 (54.4%) were HPV-positive, with the oropharynx (40,313/54,205; 74.4%) displaying the highest rates of HPV-positivity and the oral cavity (1818/11,505; 15.8%) displaying the lowest. HPV-positive malignancy was associated with significantly improved adjusted overall survival in the oropharynx (HR = 0.42 [95% CI: 0.43-0.47]), oral cavity (HR = 0.86 [95% CI: 0.79-0.95]), sinonasal tract (HR = 0.63 [95% CI: 0.48-0.83]), larynx (HR = 0.78 [95% CI: 0.71-0.87]), and hypopharynx (HR = 0.56 [95% CI: 0.48-0.66]), but not the nasopharynx (HR = 0.93 [95% CI: 0.77-1.14]). CONCLUSION: HPV testing rates were significantly lower in non-oropharyngeal subsites. This is relevant as HPV-associated disease displayed significantly improved overall survival in both the oropharynx and four of five non-oropharyngeal subsites. While validation with prospective studies is necessary, these findings may warrant HPV testing in all HNSCC subsites.

Early feeding after free flap reconstruction of the oral cavity: A systematic review and meta-analysis.

BACKGROUND: Traditionally, patients undergoing free flap reconstruction for oral cavity defects have been given nothing by mouth for 6-14 days post-operatively due to concern for orocutaneous fistula development. METHODS: Multiple databases were screened for studies assessing the rate of orocutaneous fistula formation in early (≤5 days) versus late (>5 days) feeding groups following oral cavity free flap reconstruction. Fixed- and random-effects meta-analyses were used. RESULTS: One randomized controlled trial, one prospective cohort, and three retrospective cohort studies were included. The early feeding group displayed no significant increase in orocutaneous fistula formation (RD = -0.02, p = 0.06) or free flap failure (RD = -0.01, p = 0.39), with a significantly shorter hospital length of stay (mean difference [days] = -2.43, p < 0.01). CONCLUSIONS: While further prospective trials are necessary, initiation of oral intake before post-operative day 5 may be appropriate in properly selected patients following oral reconstruction.