Patient Comes First
AI in Surgical Outcomes and Oncology

Our lab aims to utilize Artificial Intelligence and Machine Learning (AI/ML) to improve the delivery of surgical and cancer care. While most AI/ML project in medicine are built starting with data (for example a CXR, EKG, CT scan images) or a question, our work starts, and ends, with the patient. We utilize the patient’s entire journey, including all the complex data points across the journal, to develop novel and innovative frameworks to improve outcomes. In addition to working with traditional structured and unstructured datasets we utilize medical imaging, patient-reported outcomes data, and surgical videos to optimize surgical outcomes and improve how we deliver healthcare. Our goal is not simply to use AI/ML as a tool for individual projects (aka “Surgical AI”), but rather to develop a system that integrates AI/ML directly into the delivery of an optimized healthcare system (Surgical AI2).

AI Helps Improving Surgical Outcomes

Surgery also incorporates a vast number of data; including data from imaging, videos, surgical robots, large structured and unstructured datasets, and unlike many areas in medicine has concrete defined outcomes metrics that can be optimized. On a patient level highly optimized surgery can be live saving, but if not optimized unfortunately also life-threatening. On a system level surgery is the most expensive component of many healthcare system. An optimized surgical practices often translate to financial sustainability and profitability for organizations, which allows organization to offer the spectrum of other essential services, while inefficiencies can lead to unsustainable cost. Therefore, optimizing surgical efficiencies and outcomes is essential to providing optimized patient care in the modern era. In addition to surgical outcomes our lab is also focused on oncology. Caner impacts an estimated 1.9 million new people per year in the US alone and the total cost of cancer case is estimated to be 206 billion dollars in 2023, or 1% of total GDP. Despite the increased use of multimodal therapies before and after surgery, surgery remains one of the most important options for many patients with cancer. While studying all components of cancer care is important, surgery offers the advantage of being a well-defined process that is ideal for process optimization and innovation.

  • One of our core aims is to improve the efficiency, quality, and effectiveness of the delivery of surgical care. Our goal is to harness readily available, but traditionally under-utilized, data to help design an optimize surgical practice and experience for patients. This includes using AI predictive modeling, process automation, pattern recognition, and decision support. Our work includes utilizing AI amalgamate already collected Electronic Medical Record (EMR) data to identify patients eligible for early discharge, developing recommender systems aimed at improving and automating a patient workup prior to the first touchpoint, and creating automated and effective alert systems.

  • The management of cancer in the modern era is both and art and a science. While randomized trials and standard of care treatment algorithms can guide many aspects of cancer care the process also has to be personalized to the individual patient. We aim to utilize AI to help synthesize large amounts of available literature as well as provide evidence based individualized treatments plans. Ongoing work includes predictive modeling and building systems for analyzing radiology images to identify cancer subtypes non-invasively.

  • Our team has extensive experience working with institutional and national large structured and semi-structured databases to improve surgical outcomes. As the size of these databases grows, we aim to utilize AI to supplement traditional research methodologies. This has the potential to process large amounts of data much faster than traditional mechanisms and can identify patterns that are otherwise missed.

  • Traditional surgical outcomes research has relied on structured, semi-structured, and unstructured databases. Our team aims to build on the existing mechanisms and databases available for surgical research by combining them with other large datasets like imaging, videos, semi-structured patient data that has previously not been able to be efficiently utilized. Patient-centered outcomes are important because they focus on the perspectives, preferences, and experiences of patients, which are essential to delivering high-quality care. Our team has extensive experience working with patient-centered outcomes, including patient experience and QOL data and combining this with traditional surgical outcomes measures. As the amount of data available for analysis grows traditional research methodologies are insufficient and therefore our team leverage the power of AI to improve patient-centered outcomes.

Team
Cornelius A. Thiels, D.O., M.B.A.

Clinical Lead & Co-Founder

Profile

Publications

HS
Hojjat Salehinejad, Ph.D., SMIEEE

AI Lead & Co-Founder

Profile

Publications

HS
Ashok Choudhary, Ph.D.

AI Post-Doctoroal Fellow

Profile

Publications

Hala Muaddi, M.D., Ph.D., FRCSC

Hepatopancreatobiliary Fellow

Profile

Publications

Chi Zhang, M.D.

Surgery Resident

Profile

Publications

Ahmer Sultan, M.D.

Research Trainee

Profile

Publications

Samer A. Abdulmoneim, M.D.

Research Trainee

Profile

Publications

Media
Hepatic artery infusion pump chemotherapy: Surgical treatment for colorectal liver metastases - Mayo Clinic Comprehensive Cancer Center
Tramadol study - Mayo Clinic
Opioid Prescription Guidelines After Surgery - Cancer Network
There really is no safe opioid': Study finds tramadol isn't a less addictive painkiller - CBC News
Minimally invasive and complex multivisceral surgeries for patients who traditionally have not been candidates for surgery - Mayo Clinic







Openings

Postdoctoral Fellow

Location: Mayo Clinic, Rochester, MN