How to Sell Machine Learning Algorithms to Healthcare Providers

Publication Date: 30/03/2017

Written by

Simon Harris

One of the greatest commercial challenges for developers of medical image analysis algorithms is how to take their products to market. Most independent software vendors (ISVs) of image analysis solutions only offer a handful of algorithms for specific use-cases, e.g. coronary calcium scoring, bone age prediction, detection of lung nodules, etc. However, most generalist radiologists require a comprehensive “analytical tool kit” with a broad portfolio of algorithms that can detect a wide range of conditions for multiple body sites and across multiple modalities. Locating, evaluating and sourcing image analysis algorithms on a piecemeal basis from multiple vendors will be a cumbersome and time consuming process for healthcare providers. Not to mention the challenges associated with integrating the algorithms with the providers’ existing healthcare IT infrastructure. Whilst this may be a viable option for the larger academic hospitals and IDNs, most providers will not have the necessary resources for this and instead will prefer to deal with a small number of vendors, and ideally a single supplier.

There are several routes to market for image analysis ISVs, as follows:

  1. Develop an in-house image analysis workstation or platform (proprietary or open)
  2. Partner with established imaging IT companies, e.g. PACS, viewer and advanced visualisation companies
  3. Partner with modality companies
  4. Partner with healthcare ecosystem (open platform) providers
  5. Partner with companies who provide vendor agnostic image analysis platforms

The advantages and disadvantages of each, as viewed through the lens of algorithm developers, are presented below.

1. Develop an in-house image analysis platform

Examples: iCAD PowerLook Advanced Mammography Platform (AMP), RADLogics AlphaPoint‚Ñ¢, HealthMyne QIDS

Advantages: A viable option for specific clinical applications, e.g. breast and lung cancer screening. Solutions can be highly customised for specific customer types, e.g. breast imaging specialists. Full control of the development roadmap.

Disadvantages: Limited choice of algorithms for general radiology. High product development, marketing and sales costs.

2. Partner with established imaging IT companies

Examples: Most of the major PACS and advanced visualisation companies offer clinical applications from third party vendors, alongside their in-house applications. For example, GE offers over 50 clinical applications for its AW advanced visualisation platform, some of which are licensed from third party developers.

Advantages: Access to an established customer base. Tight integration with partner’s imaging IT platform. Partnering with a well-known brand may add credibility by association. Leverage the partner’s sales and marketing efforts.

Disadvantages: The imaging IT market is fragmented – being tied to a specific vendor(s) gives access to only a fraction of the total available market. The Imaging IT market is evolving from departmental PACS to enterprise imaging solutions, creating uncertainty and complexity in the marketplace.

3. Partner with modality companies

Examples: Arterys has a non-exclusive, co-marketing agreement with GE Healthcare, whereby Arterys 4D Flow is available via the ViosWorks application for GE MRI scanners.

Advantages: Access to an established customer base. Credibility by association. Leverage partner’s sales and marketing efforts. Access to “raw data” direct from the modality may improve accuracy of algorithms.

Disadvantages: Doesn’t give access to the total market, although the modality markets are more consolidated than Imaging IT. For example, the MRI market is largely controlled by an oligopoly of 5 companies – Siemens, GE, Philips, Toshiba and Hitachi. Long sales cycles. Modality companies are likely to embed a small number of algorithms rather than a full suite, which will limit the available market.

4) Partner with healthcare ecosystem (open platform) providers

Examples: GE Health Cloud (features applications from Arterys, Pie Medical Imaging and imbio, to name a few), IBM Watson Health Core (recently added an application from MedyMatch that detects intracranial bleeds on CT scans), NTT DATA Unified Clinical Archive (offers analytical solutions from imbio, Zebra Medical Vision and AnatomyWorks), Siemens Healthineers Digital Ecosystem (announced at HIMSS 2017 with Arterys, SyntheticMR and a handful of others having already agreed to provide applications).

Advantages: Widest choice of algorithms. Major focus of investment by the major healthcare technology vendors (GE plans to invest $500m over the next three years in its Health Cloud platform). Access to the platform developer’s installed base of customers. Credibility by association. Leverage the platform developer’s sales and marketing efforts.

Disadvantages: Some resistance from healthcare providers to cloud-based platforms, often due to data compliance requirements. Ecosystem platforms are a relatively new and unproven concept in medical imaging and currently there are relatively few healthcare providers using them.

5) Partner with companies who provide dedicated, vendor agnostic image analysis platforms

Examples: Medimsight offers a cloud-based computer-aided diagnosis marketplace for biomarker quantification. The platform features 39 applications, including algorithms from LAIMBIO, FMRIB (Oxford Centre for Functional MRI of the Brain) and Martinos Center for Biomedical Imaging. Blackford Analysis offers a vendor-neutral pre-processing (VNP) platform that acts as a broker for pre-processing algorithmic solutions from third party developers, to enable integration with existing clinician workflows. McCoy Medical is a distribution partner / sales channel for companies who make algorithms and analytics.

Advantages: Support with integration reduces the need for PACS back-end engineering. A highly focused marketplace for image analysis solutions.

Disadvantages: The developers of dedicated, vendor agnostic image analysis platforms are small companies with limited resources and few customers. Strong competition from healthcare ecosystem providers (see 4 above).

The Signify View

In the short-term we expect image analysis ISVs to focus on developing their own platforms and to seek partnerships with established imaging IT vendors. However, with the major healthcare technology vendors investing heavily in their healthcare ecosystem platforms, these new “clinical application marketplaces” look set to be an increasingly important sales channel in the coming years. The single platform model greatly simplifies purchasing and workflow integration for healthcare providers and gives radiologists access to the widest selection of algorithms to build their “analytical tool kits”.

Related Reports

Machine Learning in Medical Imaging – 2017 Edition” provides a data-centric and global outlook on the current and projected uptake of machine learning in medical imaging. The report blends primary data collected from in-depth interviews with healthcare professionals and technology vendors, to provide a balanced and objective view of the market. If you would like further information please contact