Sandhya Kalavacherla
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Sandhya Kalavacherla has a total of 120 co-authors. Here are the top co-authors by number of shared publications:
- Theresa Guo
Moores Cancer Center, University of California San Diego, La Jolla, CA 92093, USA
4 shared publications Jacqueline J. Greene
Department of Otolaryngology- Head and Neck Surgery, University of California San Diego, La Jolla, CA, USA
3 shared publications- Jacqueline J. Greene
Department of Otolaryngology, Head & Neck Surgery, University of California San Diego, San Diego, California, USA
3 shared publications Andrew Lowy
Department of Surgery, Division of Surgical Oncology, UC San Diego School of Medicine, 3855 Health Sciences Drive, # 0987, La Jolla, San Diego, CA, 92093, USA
2 shared publicationsBrent S. Rose
Department of Radiation Medicine and Applied Sciences University of California San Diego Health La Jolla California USA
2 shared publicationsEric Du
Department of Otolaryngology- Head and Neck Surgery, University of Missouri-Columbia, Columbia, MO, USA
2 shared publications
Latest Publications
- 16 March 2026
- Published by SAGE Publications in Facial Plastic Surgery & Aesthetic Medicine
Learning from Machine Learning: Advancing from Static Images to Dynamic Video-Based Quantification of Facial Palsy
Background: An automated method to accurately quantify facial function from videos has been a long-standing challenge in facial palsy (FP) management. Objective: To compare the accuracy of a Python open-source machine learning algorithm (Python-OS) to a standard image-based analysis tool (Emotrics) to track facial movement among patients with FP, as measured by error rates. Methods: Landmarks were generated on patient with FP images using Python-OS and Emotrics and on patient videos using Python-OS. Weighted error rates were calculated and compared between algorithms using analysis of variance tests. Results: Overall major error rates were 50.3%, 54.3%, and 9.2% for the Emotrics image, Python-OS image, and Python-OS video analyses ( p < 0.001). Compared to image analyses, Python-OS video analysis had higher accuracy across all facial features ( p = 0.03) and FP severities ( p < 0.001). Video analysis allowed us to distinguish FP-specific temporal patterns; the linear relationship between right and left oral commissure movements in normal function ( R = 0.99) became nonlinear in flaccid ( R = 0.75) and synkinetic ( R = 0.72) FP. Conclusion: We report high relative accuracy of dynamic FP quantification through Python-OS, improving the clinical utility of AI-aided FP assessment.
- 16 March 2026
- Published by SAGE Publications in Facial Plastic Surgery & Aesthetic Medicine
Learning from Machine Learning: Advancing from Static Images to Dynamic Video-Based Quantification of Facial Palsy
Background: An automated method to accurately quantify facial function from videos has been a long-standing challenge in facial palsy (FP) management. Objective: To compare the accuracy of a Python open-source machine learning algorithm (Python-OS) to a standard image-based analysis tool (Emotrics) to track facial movement among patients with FP, as measured by error rates. Methods: Landmarks were generated on patient with FP images using Python-OS and Emotrics and on patient videos using Python-OS. Weighted error rates were calculated and compared between algorithms using analysis of variance tests. Results: Overall major error rates were 50.3%, 54.3%, and 9.2% for the Emotrics image, Python-OS image, and Python-OS video analyses ( p < 0.001). Compared to image analyses, Python-OS video analysis had higher accuracy across all facial features ( p = 0.03) and FP severities ( p < 0.001). Video analysis allowed us to distinguish FP-specific temporal patterns; the linear relationship between right and left oral commissure movements in normal function ( R = 0.99) became nonlinear in flaccid ( R = 0.75) and synkinetic ( R = 0.72) FP. Conclusion: We report high relative accuracy of dynamic FP quantification through Python-OS, improving the clinical utility of AI-aided FP assessment.
- 5 December 2025
- Published by S. Karger AG in ORL
- Vol. 87 (5-6) , 1-6
A Possible Expanded Timeline for Salvage Facial Reanimation Surgery: A Case Report
Introduction: Successful salvage facial reanimation procedures involving transfer of nerve to masseter (NTM) to an existing gracilis free muscle transfer (GFMT) have been reported when performed within 6 months. Methods: Medical records were reviewed for patients who had undergone a salvage or secondary facial reanimation surgery after an initial suboptimal GFMT by cross-face nerve graft (CFNG). Case Presentations: Between 2019 and 2024, three patients underwent salvage or secondary facial reanimation surgery after an initial procedure failed to produce movement by 11, 17, and 14 months respectively. Intraoperatively, the existing gracilis of patients 1 and 3 stimulated at 2 mA, and so a salvage procedure involving NTM transfer to the obturator nerve of the existing GFMT was performed. Successful smile was noted at 5 months for patient 1 and no facial movement after 12 months for patient 3. For patient 2, a secondary facial reanimation surgery (GFMT from the contralateral leg with innervation by the NTM) was performed due to lack of existing gracilis electrical stimulation and produced a successful smile noted at 7 months. Conclusion: This study demonstrates that successful salvage of a failed GFMT by CFNG is possible through transfer of the NTM to the obturator nerve even after 11 months.
- 5 December 2025
- Published by S. Karger AG in ORL
- Vol. 87 (5-6) , 1-6
A Possible Expanded Timeline for Salvage Facial Reanimation Surgery: A Case Report
Introduction: Successful salvage facial reanimation procedures involving transfer of nerve to masseter (NTM) to an existing gracilis free muscle transfer (GFMT) have been reported when performed within 6 months. Methods: Medical records were reviewed for patients who had undergone a salvage or secondary facial reanimation surgery after an initial suboptimal GFMT by cross-face nerve graft (CFNG). Case Presentations: Between 2019 and 2024, three patients underwent salvage or secondary facial reanimation surgery after an initial procedure failed to produce movement by 11, 17, and 14 months respectively. Intraoperatively, the existing gracilis of patients 1 and 3 stimulated at 2 mA, and so a salvage procedure involving NTM transfer to the obturator nerve of the existing GFMT was performed. Successful smile was noted at 5 months for patient 1 and no facial movement after 12 months for patient 3. For patient 2, a secondary facial reanimation surgery (GFMT from the contralateral leg with innervation by the NTM) was performed due to lack of existing gracilis electrical stimulation and produced a successful smile noted at 7 months. Conclusion: This study demonstrates that successful salvage of a failed GFMT by CFNG is possible through transfer of the NTM to the obturator nerve even after 11 months.
- Open Access
- 22 April 2025
- Published by Wiley
Don't Sleep on the High Prevalence of Obstructive Sleep Apnea in the Pediatric Aerodigestive Population
Objectives : To characterize obstructive sleep apnea (OSA) and its severity in a referred pediatric population undergoing polysomnography (PSG) evaluation at an aerodigestive (AD) clinic. Study design : A total of 234 AD children at a single institution were retrospectively reviewed. Demographic characteristics, symptoms, and respiratory indices were collected. Patients’ medical complexity scores (MCS), a comorbidity burden metric on a scale of 0-7, were calculated based on institution-specific criteria. Kruskal-Wallis rank sum and Fisher’s exact tests compared sociodemographic and clinical characteristics by OSA severity. Logistic multivariable regression models identified factors associated with a moderate/severe OSA diagnosis. Results : Among 234 AD patients, 86 had PSG studies (mean 1.86 studies/patient). The median age at first study was 2 years; 53% were male, 27% Non-Hispanic White, and 49% Hispanic. OSA was diagnosed in 92% of those who completed a PSG, and the OSA severity were mild in 45.3%, moderate in 26.7%, and severe in 19.8%. Median MCS was 4 with no significant differences between OSA severity groups. Female gender (OR 2.51), age >9 years (OR 3.22), and genetic/syndromic diagnoses (OR 2.44) were significantly associated with moderate/severe OSA. Multimodal treatments with pharmacologic therapy, noninvasive support, and surgical interventions were the most effective in mitigating disease progression compared to single-modality approaches (pConclusions : OSA is highly prevalent in pediatric AD patients, with increased likelihood of moderate/severe disease in females, older children, and those with genetic/syndromic diagnoses. We report no association between OSA severity and higher medical complexity. Consequently, OSA screening should be considered in all pediatric AD patients.
- 22 April 2025
- Published by Wiley
Open AccessDon't Sleep on the High Prevalence of Obstructive Sleep Apnea in the Pediatric Aerodigestive Population
Objectives : To characterize obstructive sleep apnea (OSA) and its severity in a referred pediatric population undergoing polysomnography (PSG) evaluation at an aerodigestive (AD) clinic. Study design : A total of 234 AD children at a single institution were retrospectively reviewed. Demographic characteristics, symptoms, and respiratory indices were collected. Patients’ medical complexity scores (MCS), a comorbidity burden metric on a scale of 0-7, were calculated based on institution-specific criteria. Kruskal-Wallis rank sum and Fisher’s exact tests compared sociodemographic and clinical characteristics by OSA severity. Logistic multivariable regression models identified factors associated with a moderate/severe OSA diagnosis. Results : Among 234 AD patients, 86 had PSG studies (mean 1.86 studies/patient). The median age at first study was 2 years; 53% were male, 27% Non-Hispanic White, and 49% Hispanic. OSA was diagnosed in 92% of those who completed a PSG, and the OSA severity were mild in 45.3%, moderate in 26.7%, and severe in 19.8%. Median MCS was 4 with no significant differences between OSA severity groups. Female gender (OR 2.51), age >9 years (OR 3.22), and genetic/syndromic diagnoses (OR 2.44) were significantly associated with moderate/severe OSA. Multimodal treatments with pharmacologic therapy, noninvasive support, and surgical interventions were the most effective in mitigating disease progression compared to single-modality approaches (pConclusions : OSA is highly prevalent in pediatric AD patients, with increased likelihood of moderate/severe disease in females, older children, and those with genetic/syndromic diagnoses. We report no association between OSA severity and higher medical complexity. Consequently, OSA screening should be considered in all pediatric AD patients.
- 20 February 2025
- Published by Springer Nature in Journal of Gastrointestinal Cancer
- Vol. 56 (1) , 1-10
Survival Outcomes Between Minimally Invasive and Open Gastrectomy in Early and Locally Advanced Gastric Adenocarcinoma in a Western Center
Purpose While minimally invasive gastrectomy (MIS) is well-utilized in Asia, its adoption in the West to treat gastric adenocarcinoma has been slower. We compare survival outcomes between open gastrectomy and MIS in a high-volume Western US center. Methods In this retrospective review, demographic and clinical characteristics of gastric adenocarcinoma patients who underwent curative-intent MIS (robotic or laparoscopic approaches) or open surgery were compared via descriptive statistics. Multivariable Cox hazard regression models were constructed to assess the effects of gastrectomy type on overall survival (OS) and recurrence-free survival (RFS) in the overall cohort and a locally advanced subgroup (pathologic stage 2–3 patients). Results A total of 135 gastric adenocarcinoma patients were studied; 67% underwent MIS. Open patients experienced lower lymph node retrieval (p = 0.004) and neoadjuvant chemotherapy administration (p = 0.037) than MIS. OS (p = 0.18) and RFS (p = 0.74) were not different between MIS and open over a 5-year period. In multivariable survival models, gastrectomy type was not associated with OS (open hazard ratio (HR) = 1.78, p = 0.8 (compared to MIS)) or RFS (HR = 1.46, p = 0.7), while positive nodes (HR = 21.7, p = 0.003) and pathologic stage 3 (HR = 1.6, p = 0.025) were associated with poorer OS. Within the locally advanced cohort (N = 66, 67% MIS), OS (p = 0.43) and RFS (p = 0.72) were similarly not different between MIS and open patients. Conclusions This study contributes to the growing body of evidence supporting the efficacy of MIS to manage gastric cancer within Western populations. Importantly, these data highlight the utility of MIS as a treatment option for locally advanced disease where uptake has been slowest.
- 20 February 2025
- Published by Springer Nature in Journal of Gastrointestinal Cancer
- Vol. 56 (1) , 1-10
Survival Outcomes Between Minimally Invasive and Open Gastrectomy in Early and Locally Advanced Gastric Adenocarcinoma in a Western Center
Purpose While minimally invasive gastrectomy (MIS) is well-utilized in Asia, its adoption in the West to treat gastric adenocarcinoma has been slower. We compare survival outcomes between open gastrectomy and MIS in a high-volume Western US center. Methods In this retrospective review, demographic and clinical characteristics of gastric adenocarcinoma patients who underwent curative-intent MIS (robotic or laparoscopic approaches) or open surgery were compared via descriptive statistics. Multivariable Cox hazard regression models were constructed to assess the effects of gastrectomy type on overall survival (OS) and recurrence-free survival (RFS) in the overall cohort and a locally advanced subgroup (pathologic stage 2–3 patients). Results A total of 135 gastric adenocarcinoma patients were studied; 67% underwent MIS. Open patients experienced lower lymph node retrieval (p = 0.004) and neoadjuvant chemotherapy administration (p = 0.037) than MIS. OS (p = 0.18) and RFS (p = 0.74) were not different between MIS and open over a 5-year period. In multivariable survival models, gastrectomy type was not associated with OS (open hazard ratio (HR) = 1.78, p = 0.8 (compared to MIS)) or RFS (HR = 1.46, p = 0.7), while positive nodes (HR = 21.7, p = 0.003) and pathologic stage 3 (HR = 1.6, p = 0.025) were associated with poorer OS. Within the locally advanced cohort (N = 66, 67% MIS), OS (p = 0.43) and RFS (p = 0.72) were similarly not different between MIS and open patients. Conclusions This study contributes to the growing body of evidence supporting the efficacy of MIS to manage gastric cancer within Western populations. Importantly, these data highlight the utility of MIS as a treatment option for locally advanced disease where uptake has been slowest.
- Open Access
- 6 August 2024
- Published by Wiley in Laryngoscope Investigative Otolaryngology
- Vol. 9 (4) , e1317
Factors associated with total laryngectomy following organ‐preserving treatment of laryngeal SCC
Objective(s): A subset of laryngeal squamous cell carcinoma (LSCC) patients undergoing larynx preserving treatment ultimately require total laryngectomy (TL) for oncologic or functional reasons. This study aims to identify TL risk factors in these patients.Methods: Retrospective cohort study using Veterans Affairs (VA) database. T1–T4 LSCC cases treated with primary radiotherapy (XRT) or chemoradiotherapy (CRT) were assessed for TL and recurrence. Binary logistic and Cox regression and Kaplan–Meier analyses were implemented.Results: Of 5390 cases, 863 (16.0%) underwent TL. On multivariable analysis, age (adjusted odds ratio: 0.97 [0.96–0.98]; p < .001) and N3 disease (0.42 [0.18–1.00]; p = .050) were associated with reduced risk of TL, whereas current alcohol use (1.22 [1.04–1.43]; p = .015) and >T1 disease (T2, 1.76 [1.44–2.17]; p < .001; T3, 2.06 [1.58–2.68]; p < .001; T4, 1.79 [1.26–2.53]; p = .001) were associated with increased risk of TL. However, N2 (adjusted hazard ratio: 1.30 [1.10–1.55]; p = .003) and N3 (2.02 [1.25–3.26]; p = .004) disease were associated with an increased risk for local recurrence. Compared to XRT, treatment with CRT was associated with reduced risk for local recurrence after adjusting for other factors (0.84 [0.70–0.99]; p = .044). Those who do not receive TL following local recurrence have poorer disease‐specific survival (log‐rank, p < .001). In patients without local recurrence, N2 disease was associated with a fourfold increase in risk of TL (4.24 [1.83–9.82]; p < .001).Conclusion: Advanced nodal stage was associated with reduced rates of salvage TL in the setting of local recurrence, and subsequent worse prognosis after recurrence. Conversely, advanced nodal stage may increase the risk for functional salvage TL in patients without recurrence.Level of Evidence: Level 3.
- 6 August 2024
- Published by Wiley in Laryngoscope Investigative Otolaryngology
- Vol. 9 (4) , e1317
Open AccessFactors associated with total laryngectomy following organ‐preserving treatment of laryngeal SCC
Objective(s): A subset of laryngeal squamous cell carcinoma (LSCC) patients undergoing larynx preserving treatment ultimately require total laryngectomy (TL) for oncologic or functional reasons. This study aims to identify TL risk factors in these patients.Methods: Retrospective cohort study using Veterans Affairs (VA) database. T1–T4 LSCC cases treated with primary radiotherapy (XRT) or chemoradiotherapy (CRT) were assessed for TL and recurrence. Binary logistic and Cox regression and Kaplan–Meier analyses were implemented.Results: Of 5390 cases, 863 (16.0%) underwent TL. On multivariable analysis, age (adjusted odds ratio: 0.97 [0.96–0.98]; p < .001) and N3 disease (0.42 [0.18–1.00]; p = .050) were associated with reduced risk of TL, whereas current alcohol use (1.22 [1.04–1.43]; p = .015) and >T1 disease (T2, 1.76 [1.44–2.17]; p < .001; T3, 2.06 [1.58–2.68]; p < .001; T4, 1.79 [1.26–2.53]; p = .001) were associated with increased risk of TL. However, N2 (adjusted hazard ratio: 1.30 [1.10–1.55]; p = .003) and N3 (2.02 [1.25–3.26]; p = .004) disease were associated with an increased risk for local recurrence. Compared to XRT, treatment with CRT was associated with reduced risk for local recurrence after adjusting for other factors (0.84 [0.70–0.99]; p = .044). Those who do not receive TL following local recurrence have poorer disease‐specific survival (log‐rank, p < .001). In patients without local recurrence, N2 disease was associated with a fourfold increase in risk of TL (4.24 [1.83–9.82]; p < .001).Conclusion: Advanced nodal stage was associated with reduced rates of salvage TL in the setting of local recurrence, and subsequent worse prognosis after recurrence. Conversely, advanced nodal stage may increase the risk for functional salvage TL in patients without recurrence.Level of Evidence: Level 3.
- 3 August 2024
- Published by Elsevier in Transplantation Proceedings
- Vol. 56 (6) , 1531-1535
Reasons for COVID-19 Vaccine Hesitancy Among Patients Listed for Solid Organ Transplants
- 3 August 2024
- Published by Elsevier in Transplantation Proceedings
- Vol. 56 (6) , 1531-1535
Reasons for COVID-19 Vaccine Hesitancy Among Patients Listed for Solid Organ Transplants
- 6 May 2024
- Published by S. Karger AG in ORL
- Vol. 86 (3-4) , 1-5
Temporal Development of Depression and Anxiety in a Large Facial Palsy Cohort
Introduction: Diagnosis with facial palsy (FP) has been linked to increased psychosocial distress and communication disorders, but limited data exist on the temporal development of depression and anxiety after diagnosis. In a large cohort of FP patients, we characterize the rates of depression and anxiety at several timepoints post-FP diagnosis. Methods: A de-identified database of all FP patients who presented to a single healthcare system over 22 years was created using Epic SlicerDicer. Demographics and comorbidities were collected and depression and anxiety diagnosis rates at three timepoints (non-inclusive lower bounds) post-FP diagnosis were examined. Results: 3,910 FP patients were identified, with a median age of 59. 56% were female and 51% were white. At 0-6, 6-12-, and 12-36-month post-FP diagnosis, 156 (4%), 58 (1.4%), and 205 (5.2%) individuals were diagnosed with depression, and 171 (4.4%), 84 (2.1%), and 237 (6.1%) were diagnosed with anxiety. At each time point, the median time between FP and depression diagnosis (2.1, 3.4, and 11.4 months) or anxiety diagnosis (2.5, 4.0, and 11.1 months) was similar. Dual depression and anxiety diagnoses were observed in 52 (1.3%), 32 (0.8%), and 122 (3.1%) patients at each time point. Compared to the overall cohort, more patients with anxiety were female (65 vs. 56%, p < 0.001) and younger (57 vs. 59, p = 0.002), and more depressed patients were Black (7.3 vs. 3.3%, p = 0.02). Conclusions: Facial palsy may lead to increased risk of depression and/or anxiety in the first year after diagnosis as demonstrated here in one of the largest FP cohorts to date. We report high rates of depression (5.5%), anxiety (6.5%), and comorbid depression and anxiety (2.1%) occurring within 1 year after FP diagnosis. Of these, the majority occurred within the first 6 months (72%, 67%, 62%, respectively). Anxiety was more common in young female patients and depression more common in Black patients, which can inform targeted mental health resources within the first 6 months post-FP diagnosis.
- 6 May 2024
- Published by S. Karger AG in ORL
- Vol. 86 (3-4) , 1-5
Temporal Development of Depression and Anxiety in a Large Facial Palsy Cohort
Introduction: Diagnosis with facial palsy (FP) has been linked to increased psychosocial distress and communication disorders, but limited data exist on the temporal development of depression and anxiety after diagnosis. In a large cohort of FP patients, we characterize the rates of depression and anxiety at several timepoints post-FP diagnosis. Methods: A de-identified database of all FP patients who presented to a single healthcare system over 22 years was created using Epic SlicerDicer. Demographics and comorbidities were collected and depression and anxiety diagnosis rates at three timepoints (non-inclusive lower bounds) post-FP diagnosis were examined. Results: 3,910 FP patients were identified, with a median age of 59. 56% were female and 51% were white. At 0-6, 6-12-, and 12-36-month post-FP diagnosis, 156 (4%), 58 (1.4%), and 205 (5.2%) individuals were diagnosed with depression, and 171 (4.4%), 84 (2.1%), and 237 (6.1%) were diagnosed with anxiety. At each time point, the median time between FP and depression diagnosis (2.1, 3.4, and 11.4 months) or anxiety diagnosis (2.5, 4.0, and 11.1 months) was similar. Dual depression and anxiety diagnoses were observed in 52 (1.3%), 32 (0.8%), and 122 (3.1%) patients at each time point. Compared to the overall cohort, more patients with anxiety were female (65 vs. 56%, p < 0.001) and younger (57 vs. 59, p = 0.002), and more depressed patients were Black (7.3 vs. 3.3%, p = 0.02). Conclusions: Facial palsy may lead to increased risk of depression and/or anxiety in the first year after diagnosis as demonstrated here in one of the largest FP cohorts to date. We report high rates of depression (5.5%), anxiety (6.5%), and comorbid depression and anxiety (2.1%) occurring within 1 year after FP diagnosis. Of these, the majority occurred within the first 6 months (72%, 67%, 62%, respectively). Anxiety was more common in young female patients and depression more common in Black patients, which can inform targeted mental health resources within the first 6 months post-FP diagnosis.
- 28 April 2024
- Published by Wiley in Otolaryngology–Head and Neck Surgery
- Vol. 171 (3) , 764-776
Sex‐Specific Survival and Treatment Delay in Oropharyngeal Squamous Cell Carcinoma
Objective: As the majority of oropharyngeal squamous cell carcinoma (OPSCC) is diagnosed in males, outcomes among females are not well‐characterized. We identify sex‐specific factors in OPSCC to refine female prognostication.Study Design: Retrospective cohort.Setting: National Cancer Database (NCDB).Methods: OPSCC cases from the 2004 to 2019 NCDB were identified. Sociodemographic, clinical, and treatment characteristics (including timing between diagnosis and treatment administration) were compared between sexes. Multivariable Cox proportional hazard regression models were constructed to characterize survival in overall and female‐only cohorts. Similar multivariable binomial logistic regression and survival models were constructed to assess odds of treatment delays and their effects on survival, respectively.Results: A total of 192,973 OPSCC patients were identified; 36,695 (19%) were female. Females had more human papillomavirus (HPV) negative, lower clinical T and N stage, and higher comorbidity disease. Females experienced lower survival in HPV negative (hazard ratio, HR = 1.11, P < .001) but not HPV‐positive disease. Females were more likely to have any treatment initiated over the median of 28 days (odds ratio, OR = 1.04, P = .014) or delays in adjuvant radiotherapy initiation over 6 weeks (OR = 1.11, P = .032). Treatment delay over 60 days (HR = 1.17, P = .016) and delay in adjuvant therapy initiation (HR = 1.24, P = .02) were associated with worse survival among females.Conclusion: In one of the largest analyses of OPSCC, females had poorer survival than males, specifically in HPV‐negative disease, despite presentation with less advanced disease. Notably, delays in any treatment initiation and adjuvant radiotherapy initiation were more likely in HPV‐negative women and associated with worse survival, highlighting potential systemic weaknesses contributing to poor prognosis among females.
- 28 April 2024
- Published by Wiley in Otolaryngology–Head and Neck Surgery
- Vol. 171 (3) , 764-776
Sex‐Specific Survival and Treatment Delay in Oropharyngeal Squamous Cell Carcinoma
Objective: As the majority of oropharyngeal squamous cell carcinoma (OPSCC) is diagnosed in males, outcomes among females are not well‐characterized. We identify sex‐specific factors in OPSCC to refine female prognostication.Study Design: Retrospective cohort.Setting: National Cancer Database (NCDB).Methods: OPSCC cases from the 2004 to 2019 NCDB were identified. Sociodemographic, clinical, and treatment characteristics (including timing between diagnosis and treatment administration) were compared between sexes. Multivariable Cox proportional hazard regression models were constructed to characterize survival in overall and female‐only cohorts. Similar multivariable binomial logistic regression and survival models were constructed to assess odds of treatment delays and their effects on survival, respectively.Results: A total of 192,973 OPSCC patients were identified; 36,695 (19%) were female. Females had more human papillomavirus (HPV) negative, lower clinical T and N stage, and higher comorbidity disease. Females experienced lower survival in HPV negative (hazard ratio, HR = 1.11, P < .001) but not HPV‐positive disease. Females were more likely to have any treatment initiated over the median of 28 days (odds ratio, OR = 1.04, P = .014) or delays in adjuvant radiotherapy initiation over 6 weeks (OR = 1.11, P = .032). Treatment delay over 60 days (HR = 1.17, P = .016) and delay in adjuvant therapy initiation (HR = 1.24, P = .02) were associated with worse survival among females.Conclusion: In one of the largest analyses of OPSCC, females had poorer survival than males, specifically in HPV‐negative disease, despite presentation with less advanced disease. Notably, delays in any treatment initiation and adjuvant radiotherapy initiation were more likely in HPV‐negative women and associated with worse survival, highlighting potential systemic weaknesses contributing to poor prognosis among females.
- Open Access
Learning from Machine Learning: advancing from static to dynamic facial function quantification
Objectives We assess an open-source Python machine learning algorithm’s efficacy in image and video analysis of facial palsy (FP) patients. Methods Images and videos of 60 patients with varying FP severities performing standard movements were obtained from MEEI Facial Palsy database. Landmarks generated on images by the open-source algorithm (adapted from OpenCV and Dlib libraries) and Emotrics (standard for two-dimensional FP analysis) were compared. Considering the human eye as the standard for accuracy, three raters marked perceived errors in each algorithm’s tracking of five facial features. Cumulative error distributions between both algorithms were compared via normalized root mean square error. FP severity and facial feature-specific error rates were compared using ANOVA tests. Raters also analyzed open-source algorithm-generated video landmarks; similar statistical comparisons between open-source algorithms’ image and video-based analyses were performed. Results Cumulative error distribution between both algorithms’ image analyses was most robust for normal function; significant discrepancies were observed in mild/moderate flaccidity and nearly-normal/complete synkinesis. Both algorithms had similar error rates across all facial features (p=0.76) and FP severities (p=0.37). In the open-source algorithm’s video analysis, mild synkinesis (24.7%) and severe flaccidity (19.7%) had the highest error rates. Comparing image and video analyses generated by the open-source algorithm, video analyses had lower error rates across all FP severities (pConclusions We report for the first time the feasibility and relative accuracy of a Python open-source algorithm for dynamic facial landmark tracking in FP videos. The demonstrated superiority of landmark tracking with videos over images can improve objective FP quantification.
Open AccessLearning from Machine Learning: advancing from static to dynamic facial function quantification
Objectives We assess an open-source Python machine learning algorithm’s efficacy in image and video analysis of facial palsy (FP) patients. Methods Images and videos of 60 patients with varying FP severities performing standard movements were obtained from MEEI Facial Palsy database. Landmarks generated on images by the open-source algorithm (adapted from OpenCV and Dlib libraries) and Emotrics (standard for two-dimensional FP analysis) were compared. Considering the human eye as the standard for accuracy, three raters marked perceived errors in each algorithm’s tracking of five facial features. Cumulative error distributions between both algorithms were compared via normalized root mean square error. FP severity and facial feature-specific error rates were compared using ANOVA tests. Raters also analyzed open-source algorithm-generated video landmarks; similar statistical comparisons between open-source algorithms’ image and video-based analyses were performed. Results Cumulative error distribution between both algorithms’ image analyses was most robust for normal function; significant discrepancies were observed in mild/moderate flaccidity and nearly-normal/complete synkinesis. Both algorithms had similar error rates across all facial features (p=0.76) and FP severities (p=0.37). In the open-source algorithm’s video analysis, mild synkinesis (24.7%) and severe flaccidity (19.7%) had the highest error rates. Comparing image and video analyses generated by the open-source algorithm, video analyses had lower error rates across all FP severities (pConclusions We report for the first time the feasibility and relative accuracy of a Python open-source algorithm for dynamic facial landmark tracking in FP videos. The demonstrated superiority of landmark tracking with videos over images can improve objective FP quantification.