Parth Kothari

Parth Kothari

Auto-generated profile, based on publicly available information

Parth Kothari

Spine Service, Rothman Orthopaedic Institute, Thomas Jefferson University, Philadelphia, PA, USA

Research Focus & Metrics

Academic Career Length

11Years (2016-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
19 April 2021 in IEEE Transactions on Intelligent Transportation Systems
Verified
Journal Article
Open access
PKParth Kothari
Parth Kothari
SKSven Kreiss
Sven Kreiss
AAAlexandre Alahi
Alexandre Alahi
Since the past few decades, human trajectory forecasting has been a field of active research owing to its numerous real-world applications: evacuation situation analysis, deployment of intelligent transport systems, traffic operations, to name a few. In this work, we cast the problem of human trajectory forecasting as learning a representation of human social interactions. Early works handcrafted this representation based on domain knowledge. However, social interactions in crowded environments are not only diverse but often subtle. Recently, deep learning methods have outperformed their handcrafted counterparts, as they learn about human-human interactions in a more generic data-driven fashion. In this work, we present an in-depth analysis of existing deep learning-based methods for modelling social interactions. We propose two domain-knowledge inspired data-driven methods to effectively capture these social interactions. To objectively compare the performance of these interaction-based forecasting models, we develop a large scale interaction-centric benchmark TrajNet ++, a significant yet missing component in the field of human trajectory forecasting. We propose novel performance metrics that evaluate the ability of a model to output socially acceptable trajectories. Experiments on TrajNet++ validate the need for our proposed metrics, and our method outperforms competitive baselines on both real-world and synthetic datasets.

18 January 2023 in IEEE Transactions on Intelligent Transportation Systems
Verified
Journal Article
Open access
PKParth Kothari
Parth Kothari
AAAlexandre Alahi
Alexandre Alahi
Human trajectory forecasting in crowds presents the challenges of modelling social interactions and outputting collision-free multimodal distribution. Following the success of Social Generative Adversarial Networks (SGAN), recent works propose various GAN-based designs to better model human motion in crowds. Despite superior performance in reducing distance-based metrics, current networks fail to output socially acceptable trajectories, as evidenced by high collisions in model predictions. To counter this, we introduce SGANv2: an improved safety-compliant SGAN architecture equipped with spatio-temporal interaction modelling and a transformer-based discriminator. The spatio-temporal modelling ability helps to learn the human social interactions better while the transformer-based discriminator design improves temporal sequence modelling. Additionally, SGANv2 utilizes the learned discriminator even at test-time via a collaborative sampling strategy that not only refines the colliding trajectories but also prevents mode collapse, a common phenomenon in GAN training. Through extensive experimentation on multiple real-world and synthetic datasets, we demonstrate the efficacy of SGANv2 to provide socially-compliant multimodal trajectories.

1 May 2016 in Spine
Verified
Journal Article
Closed access
JSJeremy Steinberger
Jeremy Steinberger
BSBranko Skovrlj
Branko Skovrlj
NLNathan J. Lee
Nathan J. Lee
PKParth Kothari
Parth Kothari
DLDante M. Leven
Dante M. Leven
SCSamuel K. Cho
Samuel K. Cho
A retrospective cohort analysis of prospectively collected data. The aim of this study was to analyze morbidity and mortality in adult patients undergoing transoral approach using a large national database. The transoral approach to the anterior skull base and atlanto-axial cervical spine provides a direct corridor to the lower clivus, C1, C2, and occasionally C3. Due to the rarity of this approach and the unfamiliar anatomy, there is potential for significant morbidity and mortality. Adult patients undergoing transoral approach to the cervical spine from 2008 to 2012 were identified by the Current Procedural Terminology (CPT) code 22548 in the ACS NSQIP database. Cases with missing preoperative information were excluded. Univariate and multivariate analyses were performed to assess associated morbidity and mortality. One hundred twenty-six patients underwent cervical spine and clival surgery via the transoral approach. There were a total of 27 (21.4%) postoperative complications with three (2.4%) mortalities. On multivariate analysis, there was an increased risk of complications with operative time >4 hours [odds ratio (OR) 7.8, 95% confidence interval (95% CI) 1.8-33.1, P = 0.0054] and total length of stay >5 days (OR 7.5, 95% CI 2.4-23.4, P = 0.0006). The transoral approach carries significant risks of morbidity and mortality. Maintaining operative time <4 hours and LOS <5 days may decrease morbidity and mortality. 4.

29 August 2018 in Global Spine Journal
Verified
Journal Article
Open access
PDF file
KPKevin Phan
Kevin Phan
WRWilliam Ranson
William Ranson
SWSamuel J. W. White
Samuel J. W. White
ZCZoe B. Cheung
Zoe B. Cheung
JKJun Kim
Jun Kim
SCSamuel K. Cho
Samuel K. Cho
Retrospective study. To determine the rates of early postoperative mortality and morbidity in adults with hypoalbuminemia undergoing elective posterior lumbar fusion (PLF). The American College of Surgeons National Surgical Quality Improvement Program (ACS NSQIP) was examined from 2005 to 2012. Current Procedural Terminology (CPT) codes were used to query the database for adults (≥18 years) who underwent PLF and/or posterior/transforaminal lumbar interbody fusion (PLIF/TLIF). Patients were divided into those with normal albumin concentration (≥3.5g/dL) and those with hypoalbuminemia (<3.5 g/dL). Both univariate and multivariate analyses were performed. A total of 2410 patients were included, of whom 2251 (93.4%) were normoalbuminemic and 159 (6.6%) were hypoalbuminemic. Patients with preoperative serum albumin levels <3.5 g/dL were older with a higher American Society of Anesthesiologists (ASA) score, and more comorbidities, including anemia, diabetes, dependent functional status, and preoperative history of chronic steroid therapy. Hypoalbuminemic patients had higher rates of any 30-day perioperative complication (P < .001), unplanned readmission (P = .019), and prolonged length of stay (LOS) >5 days (P < .001). However, hypoalbuminemia was not significantly associated with any specific perioperative complication. On multivariate analysis, preoperative hypoalbuminemia was found to be an independent predictor of prolonged LOS (OR 2.4, 95% CI 1.7-3.5; P < .001) and unplanned readmission (OR 2.7, 95% CI 1.1-6.3; P = .023). Hypoalbuminemia was found to be an important predictor of patient outcomes in this population. This study suggests that clinicians should consider nutritional screening and optimization as part of the preoperative risk assessment algorithm. III

30 April 2023 in Asian Spine Journal
Verified
Journal Article
Open access
PDF file
BKBrian Karamian
Brian Karamian
PKParth Kothari
Parth Kothari
GTGregory Toci
Gregory Toci
MLMark J Lambrechts
Mark J Lambrechts
JCJose Canseco
Jose Canseco
GSGregory Schroeder
Gregory Schroeder
Study Design: Single-center retrospective cohort.Purpose: To compare surgical outcomes of patients based on lumbar drain variables relating to output and duration. Overview of Literature: The use of drains following lumbar spine surgery, specifically with respect to hospital readmission, postoperative hematoma, postoperative anemia, and surgical site infections, has been controversial.Methods: Patients aged ≥18 years who underwent lumbar fusion with a postoperative drain between 2017 and 2020 were included and grouped based on hospital readmission status, last 8-hour drain output (<40 mL cutoff), or drain duration (2 days cutoff). Total output of all drains, total output of the primary drain, drain duration in days, drain output per day, last 8-hour output, penultimate 8-hour output, and last 8-hour delta (last 8-hour output subtracted by penultimate 8-hour output) were collected. Continuous and categorical data were compared between groups. Multivariate logistic regression analysis and receiver operating characteristic (ROC) analysis were performed to determine whether drain variables can predict hospital readmission, postoperative blood transfusions, and postoperative anemia. Alpha was 0.05.Results: Our cohort consisted of 1,166 patients with 111 (9.5%) hospital readmissions. Results of regression analysis did not identify any of the drain variables as independent predictors of hospital readmission, postoperative blood transfusion, or postoperative anemia. ROC analysis demonstrated the drain variables to be poor predictors of hospital readmission, with the highest area under curve of 0.524 (drain duration), corresponding to a sensitivity of 61.3% and specificity of 49.9%.Conclusions: Drain output or duration did not affect readmission rates following lumbar spine surgery.

20 January 2022 in Foot & Ankle Orthopaedics
Verified
Journal Article
Open access
TMTara Moncman
Tara Moncman
DCDaniel O. Corr
Daniel O. Corr
PKParth Kothari
Parth Kothari
JOJoseph T. O'NEil
Joseph T. O'NEil
DPDavid I. Pedowitz
David I. Pedowitz
Category: Midfoot/Forefoot; Ankle Introduction/Purpose: Talonavicular arthrodesis is a commonly performed procedure in orthopedic foot and ankle surgery, for conditions including osteoarthritis, instability, and others. The surgery is frequently performed as part of a double or triple arthrodesis. Multiple surgical constructs have been described, including screws, plate/screw constructs, and staple/screw constructs. Few studies exist that specifically focus on the talonavicular joint, and those that do are often limited by low patient volumes. Studies that do examine non-union of the talonavicular joint do not offer comparative data on non-union rates between the different methods of fixation. The purpose of this study was to determine the rate of arthrodesis in talonavicular fusion surgery among different fixation techniques and determine if hardware has a significant effect on union rate among patient factors. Methods: This is a retrospective single institution, multi-surgeon study evaluating arthrodesis rates in the talonavicular joint. Adult patients undergoing primary talonavicular fusion (isolated or as part of double/triple arthrodesis) from 2015-2018 with a minimum of 3-month radiographic follow-up were studied. Patients undergoing revision talonavicular fusion or fusion as part of Charcot reconstruction were excluded. Patient radiographic and chart notes were reviewed to assess fixation type (single screw, 2-hole compression plate, or screw + staple), the status of the arthrodesis at >=3 months and any hardware complications or reoperations following the index procedure. Descriptive statistics were generated for standard patient demographics and co- morbidities. Non-union rates were calculated for the cohort as a whole and for each fixation technique utilized along with the frequency of hardware complications/removal. Univariate analysis was performed to determine risk factors for increased non- union rate, including patient demographics, co-morbidities, and fixation technique. Results: A total of 198 patients were included. Hardware included 131 double straight screws, 42 two-hole compression plates, and 25 screw + staple constructs. Seventeen patients (8.59%) had a nonunion of the talonavicular joint at >3 months, while 18 patients (9.09%) had hardware complications of breakage or prominence and 11 (5.56%) required eventual removal. Hardware type was not a significant predictor of nonunion, as 11 patients had straight screw hardware (8.40%), while 3 had a two-hole plate (7.14%) and 3 had screw + staple (12%) (p=0.801). Hardware breakage or backing out occurred in 6/131 patients with straight screws (4.58%), 5/42 with 2-hole plates (11.90%), and 7/25 with staple + screw (28%) (p=0.001). Hardware removal was necessary in 5 (3.82%) straight screw patients, 3 (7.14%) two-hole plate patients, and 3 (12%) screw + staple patients (p=0.186). Patient age, sex, diabetic and smoking status were not significant for nonunion or hardware complication. Conclusion: The type of hardware used in talonavicular arthrodesis is not a significant factor in achieving union. Staple + screw constructs may be more susceptible to hardware breakage or migration, but was not significant for the necessity of hardware removal, as a number of patients remain painless and achieve the intended union. The authors recommend that physicians can make decisions of which to utilize based on familiarity and preference. Further study is indicated on even larger patient cohorts with increased population of nonunion cases so that further significant patient risk factors for nonunion of the joint can be identified.