In many settings, machine learning models may be used to inform decisions that impact individuals or entities who interact with the model. Such entities, or agents, may game model decisions by manipulating their inputs to the model to obtain better outcomes and maximize some utility. We consider a multi-agent setting where the goal is to identify the "worst offenders:" agents that are gaming most aggressively. However, identifying such agents is difficult without being able to evaluate their utility function. Thus, we introduce a framework featuring a gaming deterrence parameter, a scalar that quantifies an agent's (un)willingness to game. We show that this gaming parameter is only partially identifiable. By recasting the problem as a causal effect estimation problem where different agents represent different "treatments," we prove that a ranking of all agents by their gaming parameters is identifiable. We present empirical results in a synthetic data study validating the usage of causal effect estimation for gaming detection and show in a case study of diagnosis coding behavior in the U.S. that our approach highlights features associated with gaming.

For each prediction task, we select features that have a causal influence on the target of prediction.

In consequential domains, it is often impossible to compel individuals to take treatment, so that optimal treatment assignments are merely suggestions when humans make the final treatment decisions. On the other hand, there can be different heterogeneity in both the actual response to treatment and final treatment decisions given recommendations. For example, in social services, a persistent puzzle is the gap in take-up of beneficial services among those who may benefit from them the most. When decision-makers have equity-for fairness-minded preferences over both access and average outcomes, the optimal decision rule changes due to these differing heterogeneity patterns. We study identification and improved/robust estimation under potential violations of positivity. We consider fairness constraints such as demographic parity in treatment take-up, and other constraints, via constrained optimization. We develop a two-stage, online learning-based algorithm for solving over parametrized policy classes under general constraints to obtain variance-sensitive regret bounds. Our framework can be extended to handle algorithmic recommendations under an often-reasonable covariate-conditional exclusion restriction, using our robustness checks for lack of positivity in the recommendation.

Drugs For what purpose was the dataset created? that do not have a current FDA-approved label UniTox was created as a unified toxicity dataset (e.g., withdrawn or discontinued drugs) are not across eight types of drug toxicities Each instance is a single drug. For each We generated information across all toxicities for instance, there are eight toxicities, and for each the same set of 2,418 drugs with the same toxicity, there is an LLM-generated summary of methodology of applying LLMs. For each drug, the relevant sections of the drug label, a ternary for each toxicity, we provide an LLM-generated prediction (No/Less/Most), and a binary summary of the relevant portions of the drug prediction (No/Yes). Each instance also provides label, as well as ternary (No/Less/Most) the unique SPL ID, allowing users to find the predictions and binary (No/Yes) predictions for exact text used to generate the instance data. Is there a label or target associated with each Who created the dataset (e.g., which team, instance?

Federated or multi-site studies have distinct advantages over single-site studies, including increased generalizability, the ability to study underrepresented populations, and the opportunity to study rare exposures and outcomes. However, these studies are complicated by the need to preserve the privacy of each individual's data, heterogeneity in their covariate distributions, and different data structures between sites. We propose a novel federated approach to derive valid causal inferences for a target population using multi-site data. We adjust for covariate shift and accommodate covariate mismatch between sites by developing a multiply-robust and privacy-preserving nuisance function estimation approach. Our methodology incorporates transfer learning to estimate ensemble weights to combine information from source sites. We show that these learned weights are efficient and optimal under different scenarios. We showcase the finite sample advantages of our approach in terms of efficiency and robustness compared to existing state-of-the-art approaches. We apply our approach to study the treatment effect of percutaneous coronary intervention (PCI) on the duration of hospitalization for patients experiencing acute myocardial infarction (AMI) with data from the Centers for Medicare & Medicaid Services (CMS).

Drones could be used for NHS-related missions in remote areas, inspecting offshore wind turbines and supplying oil rigs by 2026 as part of a new regulatory regime in the UK. David Willetts, the head of a new government unit helping to deploy new technologies in Britain, said there were obvious situations where drones could be used if the changes go ahead next year. Ministers announced plans this month to allow drones to fly long distances without their operators seeing them. Drones cannot be flown "beyond visual line of sight" under current regulations, making their use for lengthy journeys impossible. In an interview with the Guardian, Lord Willetts, chair of the Regulatory Innovation Office (RIO), said the changes could come as soon as 2026, but that they would apply in "atypical" aviation environments at first, which would mean remote areas and over open water. Referring to the NHS, Willetts said there was potentially a huge market for drone operators.

Every day, the world seems to be slipping further and further into dystopia, with President Donald Trump placing tariffs on islands inhabited by penguins and the country's head of Medicare and Medicaid touting AI-first healthcare. In case you needed an even higher dose of Orwellian anxiety in your life, though, Black Mirror has finally returned for season 7 with six brand new episodes. In its new season, the anthology series about our, shall we say, complicated relationship with technology takes on AI sentience, subscription pricing models, lost loves, high school grudges, and the privatization of health care. It's also got plenty of action, romance, and a heaping helping of tech-era terror. As with any anthology series, Black Mirror has plenty of hits, and also its share of misses, and season 7 is no exception, which only makes it more perfect for ranking.