Table 3. Aims, lessons learned and recommendations regarding analysis of usage-logs generated from the presented analysis.

Set	Aim	Lessons learned	Recommendations
1	To suggest and test a way of grouping log-data based on theories of human behaviour, to improve upon the tradition of summative analysis.	By grouping usage logs into “registrations” and “navigations” we were able to more easily and meaningfully identify how patients change their interactions with the mHealth devices.	When combined with traditional measures, established theories from complementary science fields, e.g. psychology, should be used to provide additional insight for mHealth intervention studies.
2	To explore what log-data can tell us about patients’ experience or relationship with the intervention technologies.	• The reduction in usage after the first month demonstrated the “novelty effect” of this technology. • Sustainable use, past the novelty effect, are dependent on relevant and easy-to-use functions.	• Analysis should consider and account for the “novelty effect” as a “run in” period, during which patients become more familiar with a technology before the intervention begins. • Automated functionalities, e.g. automatic registration of physical activity via Bluetooth from a wearable sensor, should be incorporated into the intervention when possible.
3	To suggest how researchers can tailor administration of the intervention to patients’ preferred use of the mHealth technologies.	The cluster analysis demonstrated that individuals indeed use mHealth tools differently based on the focus, or own priorities, of their self-management.	Reminders or recommendations for continued use and self-management practice can be tailored based on usage patterns of each patient during the first 3-months.
4	To propose a solution to achieve adequate data-collection.	The variability both within and between participants’ use was expected, and can be seen as a realistic representation of self-management amongst those with Type 2 diabetes.	Suggest minimum mHealth usage requirements for intervention studies to make data collection more consistent and reliable.
5	To determine how research and analysis can approach patient collected health measures.	Self-collected health data, such as BG values, diet and exercise, can supplement health measures collected at the point-of-care by providing details of health change between consultations. However, consistency and reliability of the data is required.	While lifestyle measures such as diet and exercise can be episodic and without schedule, measures such as SMBG should be done on a consistent schedule to ensure their comparability over time and two other measures during interventions.
6	To determine what more is needed to understand not only what and how, but also why patients choose to self-manage.	Usage logs are a valuable resource for understanding how use of diabetes mHealth tools change during the intervention. However, why changes occurred during the intervention period were not clear.	Related and complementary questionnaires include, e.g. Patient Activation Measure [31], Health Education Impact Questionnaire [32], Patient Health Locus of Control [33], which measure motivation and patients’ intention to engage in their health, as well as the Health Care Climate Questionnaire [34], which may provide insights as to the impact of the therapeutic relationship related to not only engagement in self-care and health outcomes but also mHealth use.