Skip to main content

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

ARTIFICIAL INTELLIGENCE

A platform for continuous learning in oncology

Nature Cancer volume 2, pages 675–676 (2021)Cite this article

Subjects

Institutional clinical data repositories provide a depth and consistency of data elements that is difficult to match even with data pooled from diverse sources. A study now demonstrates how real-world data and clinical encounters can be captured in an integrated ‘data story’ to enable continuous learning and hypothesis generation in oncology.

Electronic health records are based on physicians’ notes and other data that document clinical encounters with a given patient; hence, they are not primarily intended to capture prognostic features or disease characteristics that could be used for outcome prediction and decision support. Nonetheless, clinical data repositories potentially provide unique resources for understanding the impact of health interventions for diverse patient populations. Making clinical data routinely usable for research is a daunting challenge for multiple reasons, including the following: (1) raw clinical data are not curated and their accuracy is variable; (2) relevant clinical data are usually multi-modal (consisting of radiological images and reports, laboratory results, pathology images and physician notes, among other formats), with each modality having its own standards and ontologies; and (3) the data accumulate in an unlinked manner, and often lack context. It is therefore typically up to the researcher’s acumen to curate the patient’s and data history post hoc. At present, it is far beyond the ability of automated systems to ingest and make use of such data without human curation, although a counter-example has been published in which data harmonization was not attempted1. In this issue of Nature Cancer, Morin et al. report a data-integrating framework that enables continuous learning from multimodal health data while simultaneously capturing and integrating longitudinal clinical records2. This framework, called ‘MEDomics’, yields analyzable longitudinal patient records (or MEDomics profiles) that generate a linked, comprehensive and usable multi-modal clinical data store in oncology with real-world patient cohorts, feeding automated machine-learning pipelines.

This is a preview of subscription content, access via your institution

Access options

Buy article

Get time limited or full article access on ReadCube.

$32.00

Buy

All prices are NET prices.

Fig. 1: Integrated data capture for continuous learning in oncology.

References

  1. Rajkomar, A. et al. NPJ Digit. Med. 1, 18 (2018).

    Article  Google Scholar 

  2. Morin, O. et al. Nat. Cancer https://doi.org/10.1038/s43018-021-00236-2 (2021).

  3. Pfister, D. G. et al. JAMA Oncol. 1, 1303–1310 (2015).

    Article  Google Scholar 

  4. Khozin, S. et al. Oncologist 23, 328–336 (2018).

    Article  CAS  Google Scholar 

  5. Vickers, A. J., Kent, M. & Scardino, P. T. Urology 102, 1–3 (2017).

    Article  Google Scholar 

  6. IOM Roundtable on Evidence-Based Medicine. The Learning Healthcare System: Workshop Summary (eds. Olsen, L., Aisner, D. & McGinnis, J.M.) (National Academies Press, 2011).

  7. Drozda, J., Zeringue, A., Dummitt, B., Yount, B. & Resnic, F. Health Technol. 2, e000024 (2020).

    Google Scholar 

  8. Di Maio, M., Perrone, F. & Conte, P. Oncologist 25, e746–e752 (2020).

    PubMed  Google Scholar 

  9. Resnic, F. S. & Matheny, M. E. N. Engl. J. Med. 378, 595–597 (2018).

    Article  Google Scholar 

  10. Heath, A. P. et al. Nat. Genet. 53, 257–262 (2021).

    Article  CAS  Google Scholar 

  11. Kush, R. D. et al. J. Biomed. Inform. 107, 103421 (2020).

    Article  CAS  Google Scholar 

  12. Sudlow, C. et al. PLoS Med. 12, e1001779 (2015).

    Article  Google Scholar 

  13. Miller, R. S. & Wong, J. L. Future Oncol. 14, 5–8 (2018).

    Article  CAS  Google Scholar 

  14. Castellanos, E. & Baxi, S. S. Cancer Discov. AACR 10, 490–491 (2020).

    Article  Google Scholar 

  15. All of Us Research Program Investigators. N. Engl. J. Med. 381, 668–676 (2019).

    Article  Google Scholar 

Download references

Author information

Authors and Affiliations

  1. Department of Medical Physics, Memorial Sloan Kettering Cancer Center, New York, NY, USA

    Joseph O. Deasy

  2. Memorial Sloan Kettering Cancer Center, New York, NY, USA

    Peter D. Stetson

Authors
  1. Joseph O. Deasy
    View author publications

    You can also search for this author in PubMed Google Scholar

  2. Peter D. Stetson
    View author publications

    You can also search for this author in PubMed Google Scholar

Corresponding author

Correspondence to Joseph O. Deasy.

Ethics declarations

Competing interests

The authors declare no competing interests.

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Deasy, J.O., Stetson, P.D. A platform for continuous learning in oncology. Nat Cancer 2, 675–676 (2021). https://doi.org/10.1038/s43018-021-00239-z

Download citation

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1038/s43018-021-00239-z

Search

Advanced search

Quick links

Nature Briefing: Cancer

Sign up for the Nature Briefing: Cancer newsletter — what matters in cancer research, free to your inbox weekly.

Get what matters in cancer research, free to your inbox weekly. Sign up for Nature Briefing: Cancer