Oops, you're using an old version of your browser so some of the features on this page may not be displaying properly.

MINIMAL Requirements: Google Chrome 24+Mozilla Firefox 20+Internet Explorer 11Opera 15–18Apple Safari 7SeaMonkey 2.15-2.23

Poster display session: Breast cancer - early stage, locally advanced & metastatic, CNS tumours, Developmental therapeutics, Genitourinary tumours - prostate & non-prostate, Palliative care, Psycho-oncology, Public health policy, Sarcoma, Supportive care

5322 - Double Blind concordance Study Of Colo-Rectal Cancer Treatment Recommendations Between Artificial Intelligence Advisory Programme Watson For Oncology(WFO) & Multidisciplinary Tumor Board(MDT)

Date

22 Oct 2018

Session

Poster display session: Breast cancer - early stage, locally advanced & metastatic, CNS tumours, Developmental therapeutics, Genitourinary tumours - prostate & non-prostate, Palliative care, Psycho-oncology, Public health policy, Sarcoma, Supportive care

Topics

Translational Research

Tumour Site

Colon and Rectal Cancer

Presenters

Somashekhar Sampige Prasannakumar

Citation

Annals of Oncology (2018) 29 (suppl_8): viii133-viii148. 10.1093/annonc/mdy279

Authors

S. Sampige Prasannakumar1, M. Sepulveda2, C. Rohit Kumar1, A. Rauthan3, P. Patil3, Y. Ramya1

Author affiliations

  • 1 Surgical oncology, Manipal Comprehensive Cancer Centre Manipal Hospital, 560017 - Bangalore/IN
  • 2 Clinical research, IBM Health, 32086 - Florida/US
  • 3 Medical Oncology, Manipal Comprehensive Cancer Center Manipal Hospital, 560017 - Bangalore/IN

Resources

Login to access the resources on OncologyPRO.

If you do not have an ESMO account, please create one for free.

Abstract 5322

Background

In this era of personalized medicine there is outburst of information and plethora of new treatment strategy. Data Tsunami overwhelms human cognitive capacity. Artificial intelligence is being used for information-intensive decision making. We present here an experience with this technology in cancer treatment decision support.

Methods

WFO was used to obtain treatment recommendations of cases that were previously evaluated by a MDT at a major cancer center in India between 2014 and 2016. A comparison was made between the oncology advisor's recommended treatment and that of the tumor board. Treatment concordance was defined as a tumor board recommendation falling into the oncology advisor’s categories of "recommended" or "for consideration" treatments. All non-concordant cases (n = 33) were re-presented to the tumor board in a blinded fashion in 2016 to address time of evaluation differences between the tumor board and the oncology advisor. Results are presented as the proportion of concordant cases.

Results

From 2014-2016 we had 126 colon cancers & 124 rectal cancers. Of colon 62 & 64 were non-metastatic & metastatic respectively, whereas in rectal cancer it was 93 & 31 respectively. Mean age of the patient was 55 years. The overall concordance at first analysis was 87%. At sub group analysis in colon (85% vs 77%) & rectum (97%vs81%) & Overall (92% vs78%) non-metastatic cases had higher concordance level than metastatic cases (Table1). There were 31 cases which were non-concordant that were re-challenged to MDT. After second review the overall concordance level improved from 87% to 95%.

Conclusions

Artificial intelligence treatment recommendations with Watson for Oncology showed high levels of concordance with a multidisciplinary tumor board. This cognitive computing technology holds much promise in helping oncologists make information intensive, evidence based treatment decisions.These findings are encouraging for the use of this technology. Additional investigations are needed to understand concordance in settings where cancer expertise and treatment options may differ.

Clinical trial identification

Legal entity responsible for the study

Ethics Board Committee, Manipal Hospitals Bangalore.

Funding

Has not received any funding.

Editorial Acknowledgement

Disclosure

All authors have declared no conflicts of interest.

This site uses cookies. Some of these cookies are essential, while others help us improve your experience by providing insights into how the site is being used.

For more detailed information on the cookies we use, please check our Privacy Policy.

Customise settings
  • Necessary cookies enable core functionality. The website cannot function properly without these cookies, and you can only disable them by changing your browser preferences.