eTech Insight – Will No-Code AI Drive Healthcare Adoption? - Cover

eTech Insight – Will No-Code AI Drive Healthcare Adoption?

The Problem: Tailoring AI to Specific Business Needs

AI companies are delivering functional algorithms based on models they develop for insurance, retail, banking, and healthcare. Customers find value in the AI algorithms supplied by the vendors but want the ability to adjust the NLP models used to deliver specific AI algorithms that are more focused for their business needs.

Allowing clients to create their own AI algorithms could result in thousands of AI algorithms, many of which could be remarkably similar in scope. This creates another dilemma for companies who pursue no-code AI—how do you create a governance structure around the process to ensure it is driving the outcomes and process efficiency to meet organization goals?

Additionally, organizations that pursue no-code AI will need to create education and training for the staff that will use these applications to ensure they understand the data models and structures, understand how to identify and eliminate data bias, and promote fairness.

For healthcare organizations, the issue of tailoring NLP processes to capture appropriate information is more challenging than it is in other industries. Extracting good data from text-based documents created by physicians or clinicians can be challenging if the extracted data is not correctly linked to a codified description of the data. If an organization has an NLP with a proven ontology engine specific to healthcare verbiage, it will likely increase the accuracy of the AI results. Constant training of the NLP models will help to improve the accuracy of AI results.

The Solution: Using Base AI Algorithms to Support Client Customization

A recent Fortune article identified Primer as a company that is pursuing a no-code AI solution for its clients. The approach Primer will use is allowing clients to modify a proven and pretuned AI algorithm. After 10–20 data samples have been processed by a modified algorithm, a noticeable increase in performance is noted. This approach is called active learning. Primer believes the AI algorithms produced with this approach can be tuned with 30% less data.

Primer identifies named entity recognition as a key requirement for creating strong AI algorithms. In their example, clients must identify proper nouns for the NLP to recognize. This is consistent with the NLP challenges in healthcare. As stated in the problem section, the need for a strong ontology to guide the NLP is a critical component needed for successful AI algorithm creation and training. To counter this challenge, Primer is providing analytic tools that will help clients identify training data that is incorrectly labeled and which data points were the most important to the AI algorithm’s classification decisions.

AI platforms require curated, accurate data to develop useful algorithms. This is driving healthcare organizations to share data. Mayo Clinic is sharing de-identified patient data to promote AI research in healthcare. Ascension is sharing de-identified patient data with Google. In a previous blog, we provided insights into how synthetic data will help to drive additional AI research. The availability of patient data to support AI in healthcare is becoming less of a challenge.

The Justification: Extending AI Capabilities Beyond Limited Data Scientist Resources

The need for data scientist resources to program, train, test, and manage AI platforms will likely be a limiting factor for many healthcare organizations to pursue and adopt AI solutions in the near future. The value of no-code AI in healthcare is that many physicians and clinicians have experience with data analytics that could be used to drive adoption of AI for supporting many healthcare processes. Another resource that hospitals have for supporting no-code AI is physician and nursing informaticists that are on staff at many healthcare organizations. No-code AI will ensure that the lack of availability or affordability of data scientists to support AI services will not be barriers to adoption by any organization.

The Players: A Market of Emerging Players

No-code AI is an emerging and immature market, especially for healthcare-focused solutions. A representative list of solutions for evaluation includes the following:

  • – used to find COVID-19 research papers for advanced research projects
  • ABBYY – provides AI algorithms to support RCM, SCM, and clinical processes
  • DataRobot – has an installed base of provider organizations
  • Big – promotes a quick deployment of machine learning algorithms developed from their tool

Success Factors

  1. Identify a no-code AI solution with a graphical interface that is intuitive for healthcare staff to learn and use.
  2. Define the staff and skill sets that will be allowed to use the no-code AI tool.
  3. Ensure that the no-code AI solution fits the needs of the organization for evaluating all or specific services.
  4. Create an AI governance team that is aligned with the enterprise data governance team.


No-code AI solutions are likely to significantly reduce the barriers of entry for AI use and adoption in healthcare. The ability to use operations and clinical staff who have data analytics experience and knowledge to fine-tune AI algorithms specific to service and patient population insights will eliminate the resource challenges that are associated with having data scientists to support AI platforms and projects. Allowing physicians to create their own AI algorithms will help to drive trust and adoption of these solutions.

The implementation of an AI solution without code will require healthcare organizations to develop and integrate AI governance into their other data analytics environments. No-code AI solutions could result in the development of hundreds of AI algorithms that may not be useful or are close duplicates to other algorithms. This scenario would result in no-code AI being a liability for the organization.

Tying no-code AI governance to enterprise data governance programs will support effective curation of data sets that can be used to train the AI algorithms. No-code AI will drive higher levels of AI adoption in healthcare.

Photo credit: chinnarach, Adobe Stock