We study an LLM fine-tuning task for designing reward functions for sequential resource allocation problems in public health, guided by human preferences expressed in natural language. This setting presents a challenging testbed for alignment due to complex and ambiguous objectives and limited data availability. We propose DPO-PRO, a robust fine-tuning algorithm based on Direct Preference Optimization (DPO), which accounts for uncertainty in the preference distribution using a lightweight Distributionally Robust Optimization (DRO) formulation. Unlike prior DRO-based variants, DPO-PRO focuses solely on uncertainty in preferences, avoiding unnecessary conservatism and incurring negligible computational overhead. We evaluate DPO-PRO on a real-world maternal mobile health program operated by the non-profit organization ARMMAN, as well as on standard alignment benchmarks. Experimental results demonstrate that our method consistently improves robustness to noisy preference signals compared to existing DPO variants. Moreover, DPO-PRO achieves comparable performance to prior self-reflection-based baseline for reward function design, while requiring significantly lower inference-time cost.

In this paper, we present PRIORITY2REWARD a Large Language Model (LLM) based application which incorporates health worker preferences for resource allocation planning in public health programs. LLMs are increasingly used to design reward functions based on human preferences in Reinforcement Learning problems. We focus on LLM-designed rewards for Restless Multi-Armed Bandits, a framework for allocating limited resources among agents. In the context of public health, our approach empowers grassroots health workers to tailor automated allocation decisions to community needs. We showcase a simulated application of PRIORITY2REWARD for a large-scale mobile health program in India. The tool allows health workers to enter natural language preferences and leverages LLMs to search for reward functions aligned with the preference. Our tool then dynamically showcases how the LLM generated reward function modifies the policy outcomes with respect to different demographic groups in the population. This can help inform policy implementation at a community level.

In many real-world applications of Reinforcement Learning (RL), deployed policies have varied impacts on different stakeholders, creating challenges in reaching consensus on how to effectively aggregate their preferences. Generalized $p$-means form a widely used class of social welfare functions for this purpose, with broad applications in fair resource allocation, AI alignment, and decision-making. This class includes well-known welfare functions such as Egalitarian, Nash, and Utilitarian welfare. However, selecting the appropriate social welfare function is challenging for decision-makers, as the structure and outcomes of optimal policies can be highly sensitive to the choice of $p$. To address this challenge, we study the concept of an $\alpha$-approximate portfolio in RL, a set of policies that are approximately optimal across the family of generalized $p$-means for all $p \in [-\infty, 1]$. We propose algorithms to compute such portfolios and provide theoretical guarantees on the trade-offs among approximation factor, portfolio size, and computational efficiency. Experimental results on synthetic and real-world datasets demonstrate the effectiveness of our approach in summarizing the policy space induced by varying $p$ values, empowering decision-makers to navigate this landscape more effectively.

Harnessing the wide-spread availability of cell phones, many nonprofits have launched mobile health (mHealth) programs to deliver information via voice or text to beneficiaries in underserved communities, with maternal and infant health being a key area of such mHealth programs. Unfortunately, dwindling listenership is a major challenge, requiring targeted interventions using limited resources. This paper focuses on Kilkari, the world's largest mHealth program for maternal and child care -- with over 3 million active subscribers at a time -- launched by India's Ministry of Health and Family Welfare (MoHFW) and run by the non-profit ARMMAN. We present a system called CHAHAK that aims to reduce automated dropouts as well as boost engagement with the program through the strategic allocation of interventions to beneficiaries. Past work in a similar domain has focused on a much smaller scale mHealth program and used markovian restless multiarmed bandits to optimize a single limited intervention resource. However this paper demonstrates the challenges in adopting a markovian approach in Kilkari; therefore CHAHAK instead relies on non-markovian time-series restless bandits, and optimizes a layered set of multiple interventions to improve listenership. We use real Kilkari data from the Odisha state in India to show CHAHAK's effectiveness in harnessing multiple interventions to boost listenership, benefiting marginalized communities. When deployed CHAHAK will assist the largest maternal mHealth program to date.

Restless multi-arm bandits (RMABs) are a model for sequentially allocating a limited number of resources to agents modeled as Markov Decision Processes. RMABs have applications in cellular networks, anti-poaching, and in particular, healthcare. For such high-stakes use cases, allocations are often required to treat different groups of agents (e.g., defined by sensitive attributes) fairly. In addition to the fairness challenge, agents’ transition probabilities are often unknown and need to be learned in real-world problems. Thus, group fairness in RMABs requires us to simultaneously learn transition probabilities and how much budget we allocate to each group. Overcoming this key challenge ignored by previous work, we develop a decision-focused-learning pipeline to solve equitable RMABs, using a novel budget allocation algorithm to prevent disparity between groups. Our results on both synthetic and real-world large-scale datasets demonstrate that incorporating fair planning into the learning step greatly improves equity with little sacrifice in utility.

In 2020, maternal mortality in India was estimated to be as high as 130 deaths per 100K live births, nearly twice the UN's target. To improve health outcomes, the non-profit ARMMAN sends automated voice messages to expecting and new mothers across India. However, 38% of mothers stop listening to these calls, missing critical preventative care information. To improve engagement, ARMMAN employs health workers to intervene by making service calls, but workers can only call a fraction of the 100K enrolled mothers. Partnering with ARMMAN, we model the problem of allocating limited interventions across mothers as a restless multi-armed bandit (RMAB), where the realities of large scale and model uncertainty present key new technical challenges. We address these with GROUPS, a double oracle–based algorithm for robust planning in RMABs with scalable grouped arms. Robustness over grouped arms requires several methodological advances. First, to adversarially select stochastic group dynamics, we develop a new method to optimize Whittle indices over transition probability intervals. Second, to learn group-level RMAB policy best responses to these adversarial environments, we introduce a weighted index heuristic. Third, we prove a key theoretical result that planning over grouped arms achieves the same minimax regret--optimal strategy as planning over individual arms, under a technical condition. Finally, using real-world data from ARMMAN, we show that GROUPS produces robust policies that reduce minimax regret by up to 50%, halving the number of preventable missed voice messages to connect more mothers with life-saving maternal health information.

Underserved communities face critical health challenges due to lack of access to timely and reliable information. Nongovernmental organizations are leveraging the widespread use of cellphones to combat these healthcare challenges and spread preventative awareness. The health workers at these organizations reach out individually to beneficiaries; however such programs still suffer from declining engagement. We have deployed SAHELI, a system to efficiently utilize the limited availability of health workers for improving maternal and child health in India. SAHELI uses the Restless Multiarmed Bandit (RMAB) framework to identify beneficiaries for outreach. It is the first deployed application for RMABs in public health, and is already in continuous use by our partner NGO, ARMMAN. We have already reached ~100K beneficiaries with SAHELI, and are on track to serve 1 million beneficiaries by the end of 2023. This scale and impact has been achieved through multiple innovations in the RMAB model and its development, in preparation of real world data, and in deployment practices; and through careful consideration of responsible AI practices. Specifically, in this paper, we describe our approach to learn from past data to improve the performance of SAHELI’s RMAB model, the real-world challenges faced during deployment and adoption of SAHELI, and the end-to-end pipeline.

This paper studies restless multi-armed bandit (RMAB) problems with unknown arm transition dynamics but with known correlated arm features. The goal is to learn a model to predict transition dynamics given features, where the Whittle index policy solves the RMAB problems using predicted transitions. However, prior works often learn the model by maximizing the predictive accuracy instead of final RMAB solution quality, causing a mismatch between training and evaluation objectives. To address this shortcoming, we propose a novel approach for decision-focused learning in RMAB that directly trains the predictive model to maximize the Whittle index solution quality. We present three key contributions: (i) we establish differentiability of the Whittle index policy to support decision-focused learning; (ii) we significantly improve the scalability of decision-focused learning approaches in sequential problems, specifically RMAB problems; (iii) we apply our algorithm to a previously collected dataset of maternal and child health to demonstrate its performance. Indeed, our algorithm is the first for decision-focused learning in RMAB that scales to real-world problem sizes.

The success of many healthcare programs depends on participants' adherence. We consider the problem of scheduling interventions in low resource settings (e.g., placing timely support calls from health workers) to increase adherence and/or engagement. Past works have successfully developed several classes of Restless Multi-armed Bandit (RMAB) based solutions for this problem. Nevertheless, all past RMAB approaches assume that the participants' behaviour follows the Markov property. We demonstrate significant deviations from the Markov assumption on real-world data on a maternal health awareness program from our partner NGO, ARMMAN. Moreover, we extend RMABs to continuous state spaces, a previously understudied area. To tackle the generalised non-Markovian RMAB setting we (i) model each participant's trajectory as a time-series, (ii) leverage the power of time-series forecasting models to learn complex patterns and dynamics to predict future states, and (iii) propose the Time-series Arm Ranking Index (TARI) policy, a novel algorithm that selects the RMAB arms that will benefit the most from an intervention, given our future state predictions. We evaluate our approach on both synthetic data, and a secondary analysis on real data from ARMMAN, and demonstrate significant increase in engagement compared to the SOTA, deployed Whittle index solution. This translates to 16.3 hours of additional content listened, 90.8% more engagement drops prevented, and reaching more than twice as many high dropout-risk beneficiaries.

The widespread availability of cell phones has enabled non-profits to deliver critical health information to their beneficiaries in a timely manner. This paper describes our work to assist non-profits that employ automated messaging programs to deliver timely preventive care information to beneficiaries (new and expecting mothers) during pregnancy and after delivery. Unfortunately, a key challenge in such information delivery programs is that a significant fraction of beneficiaries drop out of the program. Yet, non-profits often have limited health-worker resources (time) to place crucial service calls for live interaction with beneficiaries to prevent such engagement drops. To assist non-profits in optimizing this limited resource, we developed a Restless Multi-Armed Bandits (RMABs) system. One key technical contribution in this system is a novel clustering method of offline historical data to infer unknown RMAB parameters. Our second major contribution is evaluation of our RMAB system in collaboration with an NGO, via a real-world service quality improvement study. The study compared strategies for optimizing service calls to 23003 participants over a period of 7 weeks to reduce engagement drops. We show that the RMAB group provides statistically significant improvement over other comparison groups, reducing ~30% engagement drops. To the best of our knowledge, this is the first study demonstrating the utility of RMABs in real world public health settings. We are transitioning our RMAB system to the NGO for real-world use.

Learning to communicate is considered an essential task to develop a general AI. While recent literature in language evolution has studied emergent language through discrete or continuous message symbols, there has been little work in the emergence of writing systems in artificial agents. In this paper, we present a referential game setup with two agents, where the mode of communication is a written language system that emerges during the play. We show that the agents can learn to coordinate successfully using this mode of communication. Further, we study how the game rules affect the writing system taxonomy by proposing a consistency metric.