I'm a data scientist that's passionate about discovering hidden gems and insights in large and complex datasets for businesses and scientific research. I completed my Ph.D. on December 2017 at Rensselaer Polytechnic Institute co-advised by Drs. Suvranu De and Xavier Intes.
My thesis showed, for the first time, that machine learning can be used to classify and accurately predict surgical motor skills using brain imaging techniques.
Directly collaborate with cross-functional business leads and understanding the business need, aggregating & exploring data, building & validating predictive models, and deploying completed models.
Designed, built, and deployed a full-stack, machine-learning based web app that predicts supply chain service levels enterprise wide, with 36% higher balanced accuracy than food industry standards.
Manage three data scientists at US Foods to execute on cross-functional business initiatives across merchandising, supply chain, and marketing teams.
Developed and validated machine learning models (linear discriminant
analysis, support vector machines, and logistic regression) using brain imaging data to assess surgical motor skill proficiency.
Validated classification models to robustly (ROC - AUC = 0.94) predict motor skill levels with a 113% higher accuracy than current US Surgery Board Certification methods.
Led three multi-institutional NIH clinical studies with ~$2M of funding in collaboration with Massachusetts General Hospital, Harvard Medical School, Yale Medical School, and University at Buffalo.
Leveraged computer vision (OpenCV) and machine learning to implement a real-time medical device part inspection system for the cardiovascular diseases division.
Used data driven and analytical methods to prove value of automated part inspection system resulting in net cost reductions of ~$200K / year.
Current metrics for surgical motor skill assessment are often simplistic, subjective, and inconsistent. Yet these metrics are used for Board certification and haven't changed for decades. This thesis presents a much more accurate approach to measure surgical motor skill via non-invasive brain imaging.Read Thesis Invited Talk
Measuring motor skill proficiency is critical for the certification of highly skilled individuals in numerous fields, especially surgery. However, conventional measures use subjective metrics that often cannot distinguish between expertise levels. This work presents an advanced optical neuroimaging methodology that can classify expertise levels with an accuracy of ~94%.Read Article View Code
When is a resident ready to perform surgery? With regards to surgical skill assessment on real tissue, current metrics are very subjective and inconsistent. My work shows that non-invasive brain imaging can be used to accurately assess whether a trainee has successfully completed training when they practice on real tissue.Read Article
How do you know where to efficiently place brain imaging probes? This work objectively determines the best optical probe placement on the scalp according to each application using Monte Carlo based photon migration simulations.Read Article
Mechanical loading has been proven to increase bone formation for patients with osteoporosis or osteopenia. This study shows that maximum bone formation depends on the location and frequency of mechanical loading.Read Article
Surgical training assessment is very challenging problem. This work utilizes real-time computer vision and machine learning classifiers, based on Haar features, to track objects and surgical tools during resident training programs to improve skill assessment metrics.Read Article View Code
Researchers leveraging advanced scanning technologies found they could identify novice from experienced surgeons by analyzing brain scans taken as the physicians worked.
About AirTalk® Join KPCC's AirTalk with host Larry Mantle weekdays for lively and in-depth discussions of city news, politics, science, the arts, entertainment, and more. Call-in number: 866-893-5722 A new study published last Wednesday in the journal Science Advances shows how skilled surgical practitioners exhibit different brain activities from unskilled practitioners.
Using a brain imaging method called "functional near-infrared spectroscopy," researchers were able to see the activity of different regions of the brain while a surgeon performed a simulated surgery, and use patterns of that activity to evaluate their level of expertise. Dr. Arun Nemani developed this idea in his Ph.D research in biomedical engineering at Rensselaer Polytechnic Institute.
Transparent skull in doctor's office A new brain scan can tell the difference between good surgeons and inexperienced ones Prominent heart surgeon David Sabiston, who performed an early coronary bypass operation in the 1960s, led a surgical training program at Duke known colloquially as "Decade with Dave."
Full list of publications on Google scholar
A Nemani, M Yücel, U Kruger, D Gee, C Cooper, S Schwaitzberg, S De, X Intes
Science Advances (2018) 4 , eaat3807.
Impact factor: 11.51
A Nemani, U Kruger, C Cooper, S Schwaitzberg, X Intes, S De
Surgical endoscopy (2018), In press.
Impact factor: 3.747
A Nemani, C Cooper, X Intes, S De, S Schwaitzberg
JACS (2017) 225 , e22.
Impact factor: 5.122
A Nemani, X Intes, S De
Biomedical Optics (2014), BM3A. 36.
Impact factor: 2.859
L Zhao, T Dodge, A Nemani, H Yokota
Biomech. Model. Mechanobiol (2014) 13, 141-151
Impact factor: 3.323