Tuberculosis control, AI and medication adherence caught in the crosswinds of contemporary immigration policy

Abstract: Imagine you are an undocumented immigrant with tuberculosis (TB). A health worker offers you a free smart-phone for daily check-ins to make sure you are taking your medications, and to talk about issues such as possible adverse reactions: Should you accept? Unfortunately, recent technology-aided immigration enforcement developments mean that the rational answer would be to decline—even if individual patient health, population health, and progress towards eliminating TB is jeopardized.
2023 Centers for Disease Control and Prevention guidance deems smart-phone based adherence monitoring (vDOT) equivalent to the prior gold standard, in-person Directly Observed Therapy (DOT). However, as documented by, among others, investigative reporter Byron Tau and the New York Times’ Privacy Project, US Government agencies now access commercial location data databases to map the movements of millions of cellphones for immigration enforcement purposes, enabled by prima facie mundane applications, including weather, gaming or money transfer apps. AI algorithms can and have been, used to intercept users along their regular paths. Up to 2.2 million undocumented immigrants have active or latent TB and may, in principle, assist in their own deportation if they accept the offer of a free smart-phone. I describe the rationales and drivers underlying different types of uses of vDOT and discuss pros and cons of 5 principal options for responding to the dilemma posed by the dual nature of data generated by smart phones: setting its sensitivity aside; addressing it directly in consent forms; offering smart phones selectively; not offering any phones; and using technological cloaking solutions.

Keywords: tuberculosis, medication adherence, AI