JMIR Publications

ABSTRACT

Background:

Recently, machine learning (ML) has been transforming people's lives by enabling intelligent voice assistants, personalized support for purchase decisions, or efficient credit card fraud detection. ML holds the potential to improve medicine as well, especially with regard to diagnostics in clinics. In a world characterized by population growth, demographic change, and the global COVID-19 pandemic, ML systems offer the opportunity to make diagnostics more effective and efficient. However, despite the high potential of ML, only a few ML systems have been deployed in clinics yet, since their adoption differs significantly from the integration of prior health information technologies given the specific characteristics of ML.

Objective:

To foster the adoption of ML systems in clinics, this study aims to explore (1) what factors influence the adoption of ML systems for medical diagnostics in clinics, and to provide insight into (2) how these factors can be used to determine an ML maturity score of clinics.

Methods:

To gain more insight into the adoption process of ML systems for diagnostics in clinics, we conducted a qualitative study by interviewing 22 selected medical experts from clinics and their suppliers with profound knowledge in the field of machine learning. We employed a semi-structured interview guideline, asked open-ended questions, and transcribed the interviews verbatim. To analyze the transcripts, we used a content analysis approach based on the technological-organizational-environmental (TOE) framework and the healthcare-specific framework of non-adoption, abandonment, scale-up, spread, and sustainability (NASSS) in a first step. In a second step, the qualitative data was used to create a maturity model for ML adoption in clinics according to an established development process.

Results:

With the help of the interviews we were able to identify twelve ML-specific factors that influence the adoption success of ML systems in clinics. These factors are categorized according to seven domains that form a holistic ML adoption framework for clinics. These are: The technological, organizational, and environmental context that determine the readiness of clinics to adopt ML systems; the adopter system, condition, and value proposition that are critical to the subsequent implementation of ML systems in everyday clinical practice; as well as the stand-alone domain patient data, which was found to be overarching throughout the adoption process. In addition, we created an applicable maturity model that can help clinicians assess their current as-is-state in the ML adoption process.

Conclusions:

Many clinics still face major problems in implementing ML systems in their processes and thus benefiting from the potential of these systems. Therefore, both the ML adoption framework and the maturity model for ML systems in clinics can not only guide future research that seeks to explore the promises and challenges associated with ML systems in a medical setting, but also be a practical reference point for clinicians.