Natural Language Processing for Cartilage Lesions of the Ankle

Scientific Research Project Number: MA 2021 04
Place: Amsterdam UMC, Department of Orthopaedic Surgery

Algorithm-Based Interventions for The Individual Patient with A Cartilage Lesion of the Ankle: What Can We Learn from A CT-Driven Database?


Background

Osteochondral defects (OCD) in the ankle are pathologic lesions of the cartilage and its subchondral bone. These lesions can occur in up to 70% of acute ankle fractures and sprains2,3. Currently, evidence-based treatment of these defects is based on low-quality small-numbered clinical case series, making it impossible to individualize patient care at a high level1,4. We recently built a database including >1500 patients with such an OCD in the ankle and filled this with CT-scan images of the patients with an OCDs. In order to answer clinically and radiologically relevant questions from the field, it is our goal to apply highly sophisticated innovative algorithms to improve and individualize patient care in an evidence-based manner through a big CT-Driven database for ankle OCDs.


Teamwork

You will be working in a very enthusiastic team consisting of two experienced orthopedic surgeons (Prof. Kerkhoffs and Dr. Stufkens) and two PhD candidates. The team consists of members of the radiology department as well as our collaborative partners that we work closely together with. The team has a high-quality background in clinical and radiological research with a high output of publications over the past 5 to 10 years.


Project description

This is a fruitful collaboration between the Department of Orthopaedic Surgery of the Amsterdam UMC and the Department of Medical Informatics at the University of Amsterdam. Your role in our team will consist of three main responsibilities: (1+2) to design a good working code to help us develop algorithms for the automatic identification of parameters (such as patient demographic factors, disease, treatment, and outcome for our CT-driven database), (3) to design an application which will extract all relevant information automatically from our CT-driven database. As a result, for your internship specifically, you will answer clinically relevant questions that are yet unanswered through the application of sophisticated algorithms based on artificial intelligence procedures. Through your dedicated contribution, you will be able to help our future patients!


Natural language processing

The project is on Natural Language Processing (NLP). Through the application of NLP, we wish to build a code in collaboration with you to fill our database with important patient characteristics that are available in the different Electronic Patient Records (Epic, Norma). These characteristics consist of both demographic data (pre-treatment) and post-treatment data with examples being age, gender, BMI, level of sports/work, intensity of pain, degrees of range of motion of the ankle, efficacy of treatment, etc. The task would be to write a code, help develop an algorithm, and design an application which will extract all relevant information in an automatic, reliable, and effective manner through the application of Natural Language Processing (NLP).


Time

7 months


Requirements


- Strong analytical and quantitative skills and strong team-working spirit

- Strong interest in artificial intelligence (machine learning, natural-language processing, etc.)

- Programming experience in R/Python/etc.

- High interest of a student to answer daily clinical questions at our Orthopaedic Department with high-quality database research


Applications

If you are interested in this project and you wish to apply for a position in our multi-disciplinary clinical research team, please send an e-mail to Dr. Sjoerd Stufkens and Jari Dahmen (s.a.stufkens@amsterdamumc.nl & j.dahmen@amsterdamumc.nl) including a brief motivation letter and a CV.


References


1. Dahmen J, Lambers KTA, Reilingh ML, van Bergen CJA, Stufkens SAS, Kerkhoffs G. No superior treatment for primary osteochondral defects of the talus. Knee Surg Sports Traumatol Arthrosc 2018;26:2142-2157.

2. Hintermann B, Regazzoni P, Lampert C, Stutz G, Gachter A. Arthroscopic findings in acute fractures of the ankle. J Bone Joint Surg Br 2000;82:345-351.

3. Martijn HA, Lambers KTA, Dahmen J, Stufkens SAS, Kerkhoffs G. High incidence of (osteo)chondral lesions in ankle fractures. Knee Surg Sports Traumatol Arthrosc 2020;Aug 6; [Epub ahead of print] PMID: 32761358.

4. Rikken QGH, Kerkhoffs GMMJ. Osteochondral Lesions of the Talus: An Individualized Treatment Paradigm from the Amsterdam Perspective. Foot and Ankle Clinics 2020;in press, available online 11 December, 2020. doi:10.1016/j.fcl.2020.10.002.


Contact

Dr. Sjoerd Stufkens , Amsterdam UMC, Department of Orthopaedic Surgery, s.a.stufkens@amsterdamumc.nl
Jari Dahmen , Amsterdam UMC, Department of Orthopaedic Surgery, j.dahmen@amsterdamumc.nl