Digital Transformation leading to Cancer Care Transformation

Flatiron health are using a big data approach to improving cancer care

Digital transformation has had a tremendous impact on how medical data is stored. Indeed, as of 2014, over 83% of office-based physicians stored their medical records on electronically [1]; and the Affordable Care Act implemented various policy changes including financial penalties to ensure that Electronic Health Records (EHRs) reach full penetration as soon as possible [2]. This has created tremendous opportunity for companies focussing on providing EHRs, such as EPIC, which reported revenues of $1.77 billion in 2014 [3].

In 2016, an estimated 1,685,210 new cases of cancer will be diagnosed in the United States and 595,690 people will die from the disease [4]. One company is attempting to use this digital transformation not only to capture patient data for storage and reading – but to improve cancer outcomes based on this data. Flatiron health was founded in 2012 with just this ambition [5].

Flatiron’s unique business model

Flatiron’s business model has two critical components. Firstly, it provides an EMR for oncologists to edit and review patient oncology records; however, whilst this is necessary, it is not their key differentiated value proposition. Indeed, their core differentiated value proposition is the ability to use their EMR to gather patient data pertaining to cancer treatment, and provide insights to oncologists from this dat, as well as to sell this data to cancer centres, researchers and pharmaceutical companies who can use this data in order to optimize patient care. They also hope to go beyond just analysing outcomes data, and into linking this data with genomics data to develop higher level inferences, and to use the platform to assist in clinical trial enrolment. Their operating model enables this, in that their EMR is cloud based, and they have permission to anonymise all patient data for further analysis; and this analysis itself takes place via in house big data processing tools.

The image demonstrates a part of the interface for oncology analytics
The Exhibit above demonstrates some of the rich insights provided by Flatiron’s offering

Flatiron health therefore offers tremendous opportunity to improve the standard of care for cancer patients. Indeed, they are keen to point out that at present only 4% of patients are in clinical trials, and this is the only data that is widely analysed and available – Flatiron Health is aiming to take the data from the other 96% of patients that are treated such that their outcomes can be used to guide future research, drug development and clinical practice [6]. With over 1 in 5 oncologists in the USA currently using the platform, they look to be in a great position to be able to deliver on these promises.

Flatiron Health’s key challenge involves making sense of unstructured data

However, before Flatiron’s data will translate into having a meaningful impact on patients, they face a number of key challenges. Firstly, medical data is unstructured. Key information is often included in various pathology reports, letters, communication notes, and laboratory results with no clear structure; with the same phenomenon often described in different words, and the same metrics described using different units. Developing algorithms to make sense of such complicated unstructured data is a significant challenge which they face at the moment. In an interview, one of the two founders of the company noted that it is too difficult to change doctors’ behaviour over a meaningful timescale, so they are working on their data processing in order to get around this [7]. This pragmatism is particularly refreshing in a world where start-ups seem to be trying to (and often failing to) change the way people do things, as opposed to working to accommodate current work flows.

Other key challenges include the ever-changing regulation pertaining to holding confidential information, and the need for security in any cloud-based system holding such private information. However, these are not issues which Flatiron takes lightly, and indeed, they have built up a strong team of cyber security personnel to this effect.

Flatiron Health is certainly doing a great job in oncology, and it is clear why they have gone after oncology given the fact that it has a key segmented user population with unique challenges as well as an area of enormous commercial interest. However, Flatiron should think about how it can look at applying its approach to other medical specialities, as theoretically many of the principles it applies in oncology can be applied elsewhere in medicine.

Conclusion: Flatiron offers potential, but has much work to do before it can begin to revolutionize cancer care

Flatiron was only founded 4 years ago – whilst that might seem like a long time ago in the tech world, for healthcare folks it’s a mere child – biotech companies sometimes take decades before they have a product which reaches the market. Digital transformation of medical records has allowed Flatiron Health to carve out a niche business model, and I hope that this will one day translate to better outcomes for cancer patients.

Word count: 799 words

 

References

1 – http://dashboard.healthit.gov/quickstats/pages/physician-ehr-adoption-trends.php

2 – http://www.medicalrecords.com/physicians/electronic-medical-records-deadline

3 – http://www.beckershospitalreview.com/healthcare-information-technology/50-things-to-know-about-epic-and-judy-faulkner.html

4 – (https://www.cancer.gov/about-cancer/understanding/statistics

5 – https://www.flatiron.com

6 – http://fortune.com/2014/07/24/can-big-data-cure-cancer/

7 – https://www.youtube.com/watch?v=9ZT7P5JFBco

Previous:

Ditch Your Pipettes: Transcriptic is the Laboratory of the Future

Next:

Blizzard Entertainment – Weathering the digital transformation

Student comments on Digital Transformation leading to Cancer Care Transformation

  1. Thanks for a great discussion on Flatiron Health! I agree that there is a lot of potential, particularly in oncology, for assembling clinical databases. Although I see how the unstructured nature of medical data would pose a challenge to Flatiron, I question whether there are workarounds to this issue, and how necessary the unstructured data is for making an impact. Big data, machine learning, and sophisticated analytics provide the opportunity to see trends and patterns where humans do not. Thus I question how necessary some of the more unstructured data, such as physician notes and letters, are. Perhaps it would be better to examine the data without bias from physicians’ notes, etc. Also, lab tests are typically already provided in numerical format. And signal processing algorithms have already become vastly more accurate and could perhaps be used for scoring diagnostic images. Aggregating this data would already provide an immense impact in the research space. I think that having digitized and decoded physician reports would be a nice-to-have, but not at all necessary at this point in time for making a large impact.

    I also question how they differentiate themselves from other IT systems commonly used in hospitals, such as Epic? Does Epic not provide the functionalities that Flatiron Health does? It would be interesting to get the perspective of oncologists currently using the platform, and why they like it. This would be important to understand as they continue to scale, given the value of the data increases greatly with scale.

  2. I agree that changing human behaviour is a losing battle, but I wonder if penetration could be increased if there was a clear benefit to the physicians and their patients. The long term research benefits are clear, but that’s not a compelling case for change – can the system be adapted to offer benefits in real time?

  3. Thanks for the interesting post! I was curious how Flatiron monetizes there data today. I think there’s a tension in healthcare, which you pointed out, that the valuable data is also very private and protected. Given that, I wonder if Flatiron will choose to analyze the data for insights themselves or sell the data to other organizations (or both). While I understand the privacy concerns, I also think we are sacrificing progress by not making data sharing more fluid in healthcare.

  4. It’s interesting to read about this startup and thanks for sharing your thoughts. I am interested in data analytic applications in oncology study, particularly genetical correlations to cancer. However, looks like Flatiron’s short term development area is still in enhancing their EHR system and basic patient data collection and process. I’m not sure how much patient genetic data they have collected but looks like it is not their immediate focus in the short term.

Leave a comment