Supplemental Digital Content is Available in the Text.

Introduction In planning for the introduction of vaginal microbicides and other new antiretroviral (ARV)-based prevention products for women, an in-depth understanding of potential end-users will be critically important to inform strategies to optimize uptake and long-term adherence. User-centred private sector companies have contributed to the successful launch of many different types of products, employing methods drawn from behavioural and social sciences to shape product designs, marketing messages and communication channels. Examples of how the private sector has adapted and applied these techniques to make decisions around product messaging and targeting may be instructive for adaptation to microbicide introduction.DiscussionIn preparing to introduce a product, user-centred private sector companies employ diverse methods to understand the target population and their lifestyles, values and motivations. ReD Associates’ observational research on user behaviours in the packaged food and diabetes fields illustrates how ‘tag along’ or ‘shadowing’ techniques can identify sources of non-adherence. Another open-ended method is self-documentation, and IDEO's mammography research utilized this to uncover user motivations that extended beyond health. Mapping the user journey is a quantitative approach for outlining critical decision-making stages, and Monitor Inclusive Markets applied this framework to identify toilet design opportunities for the rural poor. Through an iterative process, these various techniques can generate hypotheses on user drop-off points, quantify where drop-off is highest and prioritize areas of further research to uncover usage barriers. Although research constraints exist, these types of user-centred techniques have helped create effective messaging, product positioning and packaging of health products as well as family planning information. These methods can be applied to microbicide acceptability testing outside of clinical trials to design microbicide marketing that enhances product usage.ConclusionsThe introduction of microbicide products presents an ideal opportunity to draw on the insights from user-centred private sector companies’ approaches, which can complement other methods that have been more commonly utilized in microbicide research to date. As microbicides move from clinical trials to real-world implementation, there will be more opportunities to combine a variety of approaches to understand end-users, which can lead to a more effective product launch and ultimately greater impact on preventing HIV infections.

Reflections from 3 global health programs using humancentered design (HCD) offer 3 categories of lessons for those considering similar approaches: n Planning while considering the needs of both traditional global health and HCD approaches n Engaging key stakeholders to build understanding, alignment, and buy-in from the outset n Applying approaches differently from the way both designers and global health actors are accustomed to working to promote long-term program sustainability and learning Key Implications nIf implemented appropriately, integrating HCD into global health programming can produce a virtuous cycle between co-creation, stakeholder buy-in, and quality of outputs. The more that programs engage stakeholders in an inclusive, participatory process, the greater their continued willingness and motivation. This in turn allows for more iteration and higher quality, better-tailored outputs that are more likely to be sustainably used and scaled. n To engender this virtuous cycle, programs that incorporate an HCD approach will need to be scoped differently than traditional global health programs (e.g., more flexible timelines; dedicated budget for implementation and capacity building, etc.). n Because stakeholders may perceive a higher risk of failure with a new approach, proponents of HCD are faced with a substantial burden of evidence to persuade actors to consider its benefits. However, traditional global health actors should consider alternative approaches to measuring HCD's contributions, including perceived end user value.

The threat of epidemics and pandemics has increased as our world has become more interconnected. Recent epidemics have highlighted the need for increased investment in preparedness and the critical role of the private sector in health system strengthening and preparedness. Our manuscript seeks to bring attention to and promote public–private collaboration in global health preparedness by discussing areas on which public and private organizations can focus their efforts to improve partnerships. It does this by expanding on themes discussed at a conference on public–private partnerships in pandemic preparedness, Ready Together. We hope that this article will encourage effective partnerships.

Background: A core objective of HIV/AIDS programming is keeping clients on treatment to improve their health outcomes and to limit spread. Machine learning and artificial intelligence can combine client, temporal, and locational attributes to identify which clients are at greatest risk of loss to follow-up (LTFU) and enable health providers to direct support interventions accordingly.Setting: The analysis was part of a project funded by U.S.President's Emergency Plan for AIDS Relief and United States Agency for International Development, Data for Implementation, and applied to data from publicly available sources (health facility data, geospatial data, and satellite imagery) and de-identified electronic medical record data on antiretroviral therapy clients in Nigeria and Mozambique. Methods:The project applied binary classification techniques using temporal cross-validation to predict the risk that patients would be LTFU. Classifiers included logistic regression, neural networks, and tree-based models.Results: Models showed strong predictive power in both settings.In Mozambique, the best-performing model, a Random Forest, achieved an area under the precision-recall curve of 0.65 compared against an underlying LTFU rate of 23%. In Nigeria, the best-performing model, a boosted tree, achieved an area under the precision-recall curve of 0.52 compared against an underlying LTFU rate of 27%. Conclusions:Machine-learned models outperformed current classification techniques and showed potential to better direct health worker resources toward patients at greatest risk of LTFU. Moreover, models performed equally across sex and age groups, supporting the model's generalizability and wider application.