Table 2.

Summary of recommendations according to topic.

Achieving rapid and efficient development	Understanding and promoting engagement	Advancing models and theories	Evaluating effectiveness	Evaluating cost-effectiveness	Ensuring regulatory, ethical, and information governance
Consider adopting methods from engineering and other data-intensive domains in the development cycle.	Specify and establish empirically what constitutes “effective engagement” for each DBCIa, that is, sufficient engagement to achieve the intended outcomes.	Use the large amounts of real-time, ecologically valid data generated by DBCIs to test and advance models and theories of behavior change.	Evaluate at all phases in the development cycle.	At every stage, including concept development, identify all the relevant future costs and benefits.	Ensure compliance with appropriate ethics or institutional review board processes.
Use Bayesian and related approaches to improve the predictive modeling capabilities of DBCIs.	Identify and develop valid and efficient combinations of objective and subjective measures to build and test multidimensional models of engagement.	Develop methods able to efficiently analyze large, complex data sets to test dynamic theoretical propositions and allow personalization of DBCIs.	Design evaluations for generalizability.	Take account of projected uptake as well as reach.	Identify and adhere to regulatory processes that may be required for digital medical devices.
Leverage advances in data science such as machine learning, but ensure that human input is retained as needed.	Develop DBCIs with a person-centered and iterative approach, using mixed methods to progressively refine the DBCI to meet user requirements.	Specify the circumstances in which a proposed mechanism of action of a DBCI will produce a targeted effect and build an ontology to organize knowledge resulting from this.	Use methods of DBCI evaluation that capitalize on their unique characteristics.	Select a modeling framework appropriate for the complexity of the projections.	Ensure compliance with national standards for data handling, sharing, and interoperability, where appropriate.
		Develop DBCIs using a modular approach.	Use features of DBCIs to optimize control and access rich data streams.	Separately evaluate societal, personal, and health care cost-effectiveness.	Provide clear and transparent information on how data from the intervention will be used and shared.
		Support interdisciplinary research collaborations and transdisciplinary thinking.	Choose comparators that minimize contamination.

^aDBCI: Digital behavior change interventions.