Pith. sign in

REVIEW 2 major objections 6 minor 1 cited by

Who's Gaming the System? A Causally-Motivated Approach for Detecting Strategic Adaptation

T0 review · 2 major / 6 minor · reviewed 2026-08-11 · deepseek-v4-flash

Pith's one-line read The paper proves that although an agent's gaming-deterrence parameter is only partially identifiable, the full ranking of agents by gaming propensity is identifiable from pairwise causal effects, and it shows causal estimators catch the…

desk verdict A genuinely useful causal framing for gaming detection, with solid experiments, but the main corollary's proof needs a careful fix before building on it. read the letter →

arxiv 2412.02000 v1 pith:NR6Q5CDE submitted 2024-12-02 cs.LG cs.AI

classification cs.LGcs.AI
keywords strategicadaptationgamingdetectioncausalinferencetreatmenteffectrankingpartialidentificationMedicareupcodingaudittargetingmulti-agentsystems
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper tries to solve the problem of identifying which agents (health insurance plans, credit scorers, ride-share drivers) are gaming a machine-learning payout model most aggressively, when the agent's utility function is unknown. It models each agent by a scalar gaming deterrence parameter that scales the cost of manipulation, and proves this parameter is only partially identifiable from observed decisions: the best one can do is a lower bound. By recasting gaming detection as causal effect estimation, where each agent is a 'treatment' and the reported decision is the outcome, the paper proves the full ranking of agents by gaming propensity is identifiable from the signs of pairwise causal effects. In synthetic data, causal estimators catch the top offenders with fewer audits than payout-only or anomaly-detection baselines, and a Medicare case study finds predicted state rankings correlate with for-profit provider prevalence, a suspected driver of upcoding (over-reporting of diagnoses).

What carries the argument

The central object is the gaming deterrence parameter λ_p, a non-negative scalar that multiplies the manipulation cost in an agent's utility-maximization problem, Δ_p(d*_p) = argmax_{d̄∈[0,1]} R(d̄) − λ_p c(d̄ − d*_p). Because λ_p is only partially identifiable, the load-bearing identity is Corollary 1's equivalence between the sign of the average causal effect of treatment agent p versus agent p' on the reported-decision outcome and the ordering of λ_p. This rests on Theorem 1's monotonicity result, which compares what each agent would do on the same ground-truth population. Under conditional exchangeability, consistency, and positivity (Assumptions 6–8), the causal effect τ(p,p') is identified as E_x[E[d_i | x, p]] − E_x[E[d_i | x, p']] and can be estimated with standard causal effect estimators, yielding pairwise comparisons that are assembled into a full agent ranking.

What would settle it

Compare the method's ranking against a gold-standard ranking from randomized audits in a population where unmeasured severity is correlated with both agent assignment and reported decisions: if any pairwise estimate's sign disagrees with the true λ order beyond the error bound of Proposition 2, the ranking claim fails for that population.

Watch

Extended reading notes

Core claim

The central claim is that gaming propensity rankings are identifiable even though individual gaming parameters are not. Under assumptions of shared rewards and costs, increasing concave rewards, strictly convex manipulation costs, and conditional exchangeability given observed covariates, Theorem 1 shows that the utility-maximizing decision rate of agent p on a given population, Δ_p(d*_p), is monotonically ordered with the gaming deterrence parameter: Δ_p(d*_p) < Δ_{p'}(d*_p) if and only if λ_p > λ_{p'}. Corollary 1 re-expresses this as a causal effect: τ(p,p') = E_x[E[d_i(p)|x]] − E_x[E[d_i(p')|x]] > 0 if and only if λ_p < λ_{p'}. Since the potential outcome means are identified from observed data under Assumptions 6–8, estimating the effect of swapping which agent is responsible for each individual yields an ordinal ranking of agents by gaming deterrence, directly actionable for targeting audits. The non-identifiability of λ_p itself is established in Proposition 1: with unknown ground-truth decision rate d*_p, only the lower bound R'(Δ_p(d*_p))/c'(Δ_p(d*_p)) ≤ λ_p < ∞ can be recovered.

Load-bearing premise

The whole ranking depends on the assumption that, after conditioning on observed patient characteristics, any difference between how two agents report on the same kind of patient is due to gaming rather than to differences in care quality or unmeasured severity.

Editorial extensions

If this is right

  • Targeted audits can be prioritized by the predicted ranking, so a fixed audit budget catches more truly gaming agents than payout-based or anomaly-detection screening.
  • The ranking is identifiable without fraud labels and without knowing agents' utility functions, as long as confounders are observed and conditional exchangeability holds.
  • A payout-only ranking can be worse than random under strong confounding (for instance, when sicker patients are enrolled in more gaming-prone plans), whereas causal estimators remain valid.
  • Anomaly-detection approaches are inherently limited for gaming detection because gamed decisions need not be outliers; causal methods exploit overlap in covariate space for counterfactual comparisons.
  • The framework extends to any multi-agent setting with a payout model, including credit scoring and ride-sharing, provided the shared-rewards, cost-convexity, and exchangeability assumptions hold.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • A practical extension the authors leave implicit: the same pairwise-effect machinery could be used to monitor shifts in gaming propensity over time, flagging agents whose estimated λ ranking changes after a payout-model update.
  • The partial-identification result suggests a concrete robustness check before acting on a ranking: perturb the adjustment set with plausibly unmeasured confounders (e.g., severity proxies) and verify that pairwise effect signs do not flip, since a violation of conditional exchangeability could invert the ranking.
  • The theory also implies a stronger detection target than ranking — certifying that a specific agent is ε-gaming given bounds on the ground-truth rate and known cost/reward derivatives — which the paper dismisses as doubtful in practice but which could be revived in settings where such bounds are defensible.
  • The Medicare case study is correlational; a sharper validation would compare predicted state rankings against external audit outcomes or natural experiments such as payout-formula changes, which would simultaneously test the exchangeability assumption.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

2 major / 6 minor

Summary. The manuscript studies detection of strategic adaptation by multiple agents. It introduces a per-agent gaming deterrence parameter λ_p in a strategic classification utility model, shows that λ_p is only partially identifiable from observed decision rates, and claims that a total ordering of agents by λ_p is identifiable via causal effect estimation, where the agent indicator is a treatment and the reported decision is the outcome (Theorem 1, Corollary 1). The paper validates the approach on synthetic data and applies it to state-level Medicare diagnosis coding, reporting correlations with for-profit provider prevalence. The main theoretical result is not fully established as written due to a gap in Corollary 1's proof and a mismatch between the aggregate utility model and the individual-level estimands used in experiments.

Significance. Strategic classification is usually studied under known or identical manipulation costs; introducing a latent per-agent deterrence parameter and asking which agents game most is practically valuable for audit targeting, and the agenda of reformulating ranking as a causal effect problem is appealing. The paper is honest about the unverifiable exchangeability assumption, and the partial-identification result is correct under the stated assumptions. The synthetic evaluation is well designed in that it varies confounding strength, compares several causal estimators against anomaly-detection baselines, and reports audit-efficiency metrics; code release is promised. However, the central ranking theorem currently rests on an unproven equality in Corollary 1, and the theory is stated at the aggregate-rate level while the simulations generate and estimate individual-level conditional decision probabilities. If these gaps are repaired, the ranking-identifiability result would be a useful contribution.

major comments (2)
  1. [§4.2, Corollary 1; Appendix B.3] The proof of Corollary 1 is not valid as written. The displayed identity E[E[d_i | p, x_i]] = E[d_i | p] holds only when the outer expectation is taken over the covariate distribution of agent p, P(x | p). Applying the “symmetric” argument to p′ then produces E[E[d_i | p′, x_i]] over P(x | p′), so τ(p,p′) reduces to the observed marginal rate difference Δ_p(d*_p) − Δ_{p′}(d*_{p′}). Theorem 1, however, requires the counterfactual contrast Δ_p(d*_p) − Δ_{p′}(d*_p), in which p′ is evaluated on p's population. Equation (5) is consistent with the counterfactual target because both sums run over agent p's observations, but the Corollary statement and its proof do not define the outer expectation over a common target distribution, and the proof's second equality does not establish E_{x|p}[E[d_i(p′) | x]] = Δ_{p′}(d*_p). This step is load-bearing: without it the ranking claim does not follow from Theorem 1.
  2. [§3, Eq. (2), Assumption 5; §5.1] The theoretical framework models each agent as choosing a single aggregate rate Δ_p(d*_p) from a scalar ground-truth rate d*_p, so that P(d_i=1|p) is constant across individuals. In contrast, the synthetic data in §5.1 generate per-observation gamed probabilities α_p(i) = arg max log(d̃) − λ_p (d̃ − d*(i))² that depend on the individual d*(i), and the causal estimators in §5.1 model E[d_i | p, x_i] as a function of x_i. If d*_p in Eq. (2) is meant to be the population mean of d*(i), then the observed marginal rate P[d_i=1|p] is not equal to Δ_p(E[d*|p]) unless the utility maximizer is linear in d*, which is not assumed. This disconnect between the aggregate theory and the individual-level estimand used in experiments means the experiments do not directly test Theorem 1 and Corollary 1 as stated. The authors should either generalize the theory to individual-level d*(x) and prove the needed monotonicity of E[d_i(p)|x] in λ_p, or adjust the simulations to the aggregate model.
minor comments (6)
  1. [§4.2, Corollary 1] The notation E_{x_i} in the definition of τ(p,p′) is ambiguous: the two terms must be integrated over the same target covariate distribution, and this distribution should be stated explicitly.
  2. [Appendix B.3] The phrase “E[E[d_i | p, x_i]] is an unbiased estimator” is imprecise: this quantity is a population conditional expectation, not an estimator; “is identified by” would be more accurate.
  3. [§4.2, Proposition 2] The condition “for all p,p′ such that inf_{p,p′} |τ(p,p′)| > ε” should be a per-pair condition (e.g., |τ(p,p′)| > ε); otherwise the quantifier is stronger than needed. There is also a typo “ˆτ(p,p)” in the statement.
  4. [Appendix B.2] Equations (22) and (23) define λ*(p) twice with identical content; one of the two displays should be removed.
  5. [Appendix C.1] There is a typo, “constnat,” that should read “constant”; Appendix B.4 also has “shedule” for “schedule.”
  6. [Figure 4] The pseudocode computes agent_i_cf − agent_j_cf; a comment stating the sign convention relative to Corollary 1's τ > 0 iff λ_p < λ_{p′} would help readers map the code to the theorem.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the ranking result is derived from explicit utility and causal assumptions, not from fitting or self-citation. The B.3 proof concern is a correctness gap, not a circular reduction.

full rationale

The paper's central claim (Theorem 1 + Corollary 1) is a derivation under explicit assumptions (Assumptions 1-8), not a fit disguised as a prediction. Theorem 1 is a comparative-statics result about the utility maximization in Eq. (2)/(4): under shared increasing concave rewards and convex costs, a larger deterrence parameter lambda_p yields a smaller optimal gamed rate on the same ground-truth population; this follows from the first-order conditions in Appendix B.2. Corollary 1 attempts to connect the causal estimand tau to the two optimal rates; the identification step E[d_i(p)|x] = E[d_i|x,p] is a standard conditional-exchangeability result (proved in Appendix B.5, with the cited [37] result derived rather than merely imported). The ranking procedure is tested on synthetic data generated from the paper's own model, which is a conventional simulation check, not circular reasoning. No load-bearing self-citation is present. One non-circular correctness concern: the proof of Corollary 1 (Appendix B.3, Eq. 26) writes E[E[d_i | p, x_i]] = E[d_i | p] = P[d_i = 1 | p] = Delta_p(d*_p); this identity is valid only if the outer expectation is over P(x|p), while the 'symmetric' case for p' would then be taken over P(x|p') and yield Delta_p'(d*_p'), not the counterfactual Delta_p'(d*_p) required by Theorem 1 and Eq. (5). If instead the outer expectation in tau is over a common covariate distribution, the first displayed equality is not the marginal P[d_i = 1 | p]. This is a proof gap or ambiguity in the estimand definition, but it is not a circular reduction: the claimed ranking is not defined in terms of the fitted tau, and no fitted parameter is renamed as a prediction. Therefore the circularity score is 0.

Assumptions & free parameters 0 free parameters · 10 assumptions · 1 invented entities

The central claim rests on a set of behavioral and causal assumptions (Assumptions 1-8) plus an unverifiable exchangeability condition. The framework introduces one latent construct, λ_p, which is the target of inference rather than a fitted parameter. No free parameters are fitted to make the derivation work.

assumptions (10)
  • domain assumption Agents are rational utility maximizers: each agent chooses a report rate solving Eq. (2).
    Load-bearing for the comparative statics in Theorem 1; the paper acknowledges rationality may fail in practice (Appendix B.4).
  • domain assumption Assumption 1: Reward R and cost c functions are shared across agents.
    Needed so that differences in behavior are attributable to λ_p; Medicare exclusion of dual-eligibles is intended to preserve shared R.
  • domain assumption Assumption 2: R is strictly increasing in the reported rate.
    Gives agents incentive to increase reported decisions.
  • domain assumption Assumption 3: c is strictly convex, minimized at 0, c(0)=0, c'(0)=0.
    Ensures a unique optimum and makes the lower bound in Proposition 1 sharp.
  • domain assumption Assumption 4: R is concave (diminishing or linear returns).
    Guarantees gaming is utility maximizing and is used in the proof of Theorem 1.
  • domain assumption Assumption 5: d*_p is a constant in [0,1] depending only on x_i.
    Lets agents be described by a single ground-truth rate; this implies patient mix is the only source of variation in appropriate diagnosis rates.
  • domain assumption Assumption 6: Conditional exchangeability, d_i(p_i) ⊥ p_i | x_i.
    The core identification condition; the paper admits it is statistically unverifiable in the Limitations section. If unmeasured confounders exist, the estimated effects do not identify the desired counterfactuals.
  • standard math Assumption 7: Consistency, d_i(p_i) = d_i.
    Standard causal inference consistency; holds by construction when the observed outcome is the potential outcome under the actual agent.
  • standard math Assumption 8: Positivity/overlap, 0 < P[p|x] < 1.
    Needed for conditional expectations to be well-defined and for estimators to be consistent; may be violated for extreme patient profiles.
  • domain assumption All covariates x_i are truthfully observed and each x_i is equally likely to be gamed.
    Stated in Section 3; requires that gaming does not alter covariates and that the agent applies the same gaming propensity to every patient.
invented entities (1)
  • Gaming deterrence parameter λ_p
    purpose: Scalar parameterizing each agent's unwillingness to game; the ranking target.
    λ is a latent construct defined through the utility model; it is never directly observed. The paper provides only partial identification and an indirect ranking, and external validation is limited to weak correlations with for-profit prevalence in an exploratory case study.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Who's Gaming the System? A Causally-Motivated Approach for Detecting Strategic Adaptation." pith.science (2026). https://pith.science/paper/NR6Q5CDE

@misc{pith2026241202000,
  author       = {Pith},
  title        = {Pith review of: Who's Gaming the System? A Causally-Motivated Approach for Detecting Strategic Adaptation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/NR6Q5CDE}},
  note         = {Machine review of arXiv:2412.02000}
}
read the original abstract

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 knowledge of their utility function. Thus, we introduce a framework in which each agent's tendency to game is parameterized via a scalar. 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.

Figures

Figures reproduced from arXiv: 2412.02000 by the authors.

Figure 1
Figure 1. Left: Two agents with gaming deterrence parameters λ1 = 30 (purple) and λ2 = 50 (blue) maximize utility (reward R - cost c) with respect to diagnosis rate. Gaming costs increase in λ(·) , and lower an agent’s optimal diagnosis rate (stars). Center: Agents’ observed decisions reflect utility-maximizing behavior. Right: A decision-maker computes a payout based on agent decisions. In this work, we study how one can ide… view at source ↗
Figure 2
Figure 2. Left: Toy dataset with observed factual outcomes di(p) and di(p ′ ). “?” denotes missing counterfactual outcomes. Right: Causal graph for gaming detection with confounders x, agent indicator p, ground truth diagnosis d ∗ , and agent decision d. ...what would do? Which agent is gaming more aggresively: or ? Causally-motivated gaming detection Given ’s observations... report diagnosis? ✔ ✘ ✔ Impute counterfactual deci… view at source ↗
Figure 3
Figure 3. Causally-motivated gaming detection. Left: First, we impute counterfactual decisions for each agent. Middle: The imputed counterfactuals yield average treatment effects (ATEs) across pairs of agents. Right: Using ATE estimates to rank agents yields a ranking of the gaming parameter λp. We show one direction of comparison across agents for simplicity. In practice, we impute decisions for both directions of comparison… view at source ↗
Figures from the paper (28 more)
Figure 4
Figure 4. Figure 4: Pseudocode for causally-motivated gaming detection. Causal effect estimators take pairs [PITH_FULL_IMAGE:figures/full_fig_p006_4.png]
Figure 5
Figure 5. Figure 5: Mean top-5 sensitivity (left) and DCG (center) across # of agents audited, and top-5 [PITH_FULL_IMAGE:figures/full_fig_p008_5.png]
Figure 6
Figure 6. Figure 6: Area under the sensitivity curve (AUSC) for causal vs. non-causal methods across levels of [PITH_FULL_IMAGE:figures/full_fig_p009_6.png]
Figure 7
Figure 7. Figure 7: AUSC of S+IPW (causal) vs. KNN (non-causal) across confounding strength with ±σ error. The advantage of S+IPW over KNN decreases as con￾founding diminishes. A payout-only approach yields worse than random ranking with sufficient confounding between covariates and agent…
Figure 8
Figure 8. Figure 8: Mean top-5 sensitivity (left) and DCG (right) across # of agents audited at mean range 0.0, [PITH_FULL_IMAGE:figures/full_fig_p020_8.png]
Figure 9
Figure 9. Figure 9: Mean top-5 sensitivity (left) and DCG (right) across # of agents audited at mean range 0.1, [PITH_FULL_IMAGE:figures/full_fig_p023_9.png]
Figure 10
Figure 10. Figure 10: Mean top-5 sensitivity (left) and DCG (right) across # of agents audited at mean range 0.2, [PITH_FULL_IMAGE:figures/full_fig_p023_10.png]
Figure 11
Figure 11. Figure 11: Mean top-5 sensitivity (left) and DCG (right) across # of agents audited at mean range 0.3, [PITH_FULL_IMAGE:figures/full_fig_p023_11.png]
Figure 12
Figure 12. Figure 12: Mean top-5 sensitivity (left) and DCG (right) across # of agents audited at mean range 0.4, [PITH_FULL_IMAGE:figures/full_fig_p023_12.png]
Figure 13
Figure 13. Figure 13: Mean top-5 sensitivity (left) and DCG (right) across # of agents audited at mean range 0.5, [PITH_FULL_IMAGE:figures/full_fig_p024_13.png]
Figure 14
Figure 14. Figure 14: Mean top-5 sensitivity (left) and DCG (right) across # of agents audited at mean range 0.6, [PITH_FULL_IMAGE:figures/full_fig_p024_14.png]
Figure 15
Figure 15. Figure 15: Mean top-5 sensitivity (left) and DCG (right) across # of agents audited at mean range 0.7, [PITH_FULL_IMAGE:figures/full_fig_p024_15.png]
Figure 16
Figure 16. Figure 16: Mean top-5 sensitivity (left) and DCG (right) across # of agents audited at mean range 0.8, [PITH_FULL_IMAGE:figures/full_fig_p024_16.png]
Figure 17
Figure 17. Figure 17: Mean top-5 sensitivity (left) and DCG (right) across # of agents audited at mean range 0.9, [PITH_FULL_IMAGE:figures/full_fig_p025_17.png]
Figure 18
Figure 18. Figure 18: Mean top-5 sensitivity (left) and DCG (right) across # of agents audited at mean range 1.0, [PITH_FULL_IMAGE:figures/full_fig_p025_18.png]
Figure 19
Figure 19. Figure 19: Area under the sensitivity curve (AUSC) for all methods tested across levels of confounding [PITH_FULL_IMAGE:figures/full_fig_p025_19.png]
Figure 20
Figure 20. Figure 20: Area under the sensitivity curve (AUSC) for all methods tested across levels of confounding [PITH_FULL_IMAGE:figures/full_fig_p025_20.png]
Figure 21
Figure 21. Figure 21: Sensitivity analysis of all causal methods tested, mean range 0.0. [PITH_FULL_IMAGE:figures/full_fig_p026_21.png]
Figure 22
Figure 22. Figure 22: Sensitivity analysis of all causal methods tested, mean range 0.1. [PITH_FULL_IMAGE:figures/full_fig_p027_22.png]
Figure 23
Figure 23. Figure 23: Sensitivity analysis of all causal methods tested, mean range 0.2. [PITH_FULL_IMAGE:figures/full_fig_p027_23.png]
Figure 24
Figure 24. Figure 24: Sensitivity analysis of all causal methods tested, mean range 0.3. [PITH_FULL_IMAGE:figures/full_fig_p027_24.png]
Figure 25
Figure 25. Figure 25: Sensitivity analysis of all causal methods tested, mean range 0.4. [PITH_FULL_IMAGE:figures/full_fig_p027_25.png]
Figure 26
Figure 26. Figure 26: Sensitivity analysis of all causal methods tested, mean range 0.5. [PITH_FULL_IMAGE:figures/full_fig_p028_26.png]
Figure 27
Figure 27. Figure 27: Sensitivity analysis of all causal methods tested, mean range 0.6. [PITH_FULL_IMAGE:figures/full_fig_p028_27.png]
Figure 28
Figure 28. Figure 28: Sensitivity analysis of all causal methods tested, mean range 0.7. [PITH_FULL_IMAGE:figures/full_fig_p028_28.png]
Figure 29
Figure 29. Figure 29: Sensitivity analysis of all causal methods tested, mean range 0.8. [PITH_FULL_IMAGE:figures/full_fig_p028_29.png]
Figure 30
Figure 30. Figure 30: Sensitivity analysis of all causal methods tested, mean range 0.9. [PITH_FULL_IMAGE:figures/full_fig_p029_30.png]
Figure 31
Figure 31. Figure 31: Sensitivity analysis of all causal methods tested, mean range 1.0. [PITH_FULL_IMAGE:figures/full_fig_p029_31.png]

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Incentive-Aware Machine Learning; Robustness, Fairness, Improvement & Causality

    cs.GT 2025-05 accept novelty 1.0 of 10

    A taxonomy and literature review of strategic classification and performative prediction, organized into robustness, fairness, and improvement/causality perspectives.

Reference graph

Works this paper leans on

64 extracted references · 48 canonical work pages · cited by 1 Pith paper

  1. [1]

    Strategic classi- fication

    Moritz Hardt, Nimrod Megiddo, Christos Papadimitriou, and Mary Wootters. Strategic classi- fication. In Proceedings of the 2016 ACM conference on innovations in theoretical computer science, pages 111–122, 2016

  2. [2]

    Risk adjustment of medicare capitation payments using the cms-hcc model

    Gregory C Pope, John Kautter, Randall P Ellis, Arlene S Ash, John Z Ayanian, Lisa I Iezzoni, Melvin J Ingber, Jesse M Levy, and John Robst. Risk adjustment of medicare capitation payments using the cms-hcc model. Health care financing review, 25(4):119, 2004

  3. [3]

    Upcoding: evidence from medicare on squishy risk adjustment

    Michael Geruso and Timothy Layton. Upcoding: evidence from medicare on squishy risk adjustment. Journal of Political Economy, 128(3):984–1026, 2020

  4. [4]

    Medicare upcoding and hospital ownership

    Elaine Silverman and Jonathan Skinner. Medicare upcoding and hospital ownership. Journal of health economics, 23(2):369–389, 2004

  5. [5]

    Assessing coding practices for gastroin- testinal surgery over time in the united states

    Faiz Gani, Joseph K Canner, and Timothy M Pawlik. Assessing coding practices for gastroin- testinal surgery over time in the united states. Surgery, 164(3):530–538, 2018

  6. [6]

    Association of insur- ance mix and diagnostic coding practices in new york state hospitals

    Kacie L Dragan, Sunita M Desai, John Billings, and Sherry A Glied. Association of insur- ance mix and diagnostic coding practices in new york state hospitals. JAMA Health Forum, 3(9):e222919–e222919, 2022

  7. [7]

    Michael E. Chernew. Report to the Congress: Medicare Payment Policy, 2022

  8. [8]

    Fy2024 justification of estimates for appropriations committees, 2024

    Centers for Medicare and Medicaid Services. Fy2024 justification of estimates for appropriations committees, 2024

Show all 64 references
  1. [9]

    Measuring coding intensity in the medicare advantage program

    Richard Kronick and W Pete Welch. Measuring coding intensity in the medicare advantage program. Medicare & Medicaid Research Review, 4(2), 2014

  2. [10]

    Note to: Medicare advantage organizations, prescrip- tion drug plan sponsors, and other interested parties, 2023

    Center for Medicare & Medicaid Center. Note to: Medicare advantage organizations, prescrip- tion drug plan sponsors, and other interested parties, 2023

  3. [11]

    Pre-submission risk adjustment audits: Preventing medicare advantage plans from draining medicare funds dry

    Casie C Rodenberger. Pre-submission risk adjustment audits: Preventing medicare advantage plans from draining medicare funds dry. Iowa L. Rev., 103:841, 2017

  4. [12]

    Monitoring for waste: Evidence from medicare audits

    Maggie Shi. Monitoring for waste: Evidence from medicare audits. The Quarterly Journal of Economics, 139(2):993–1049, 2024

  5. [13]

    The scored society: Due process for automated predictions

    Danielle Keats Citron and Frank Pasquale. The scored society: Due process for automated predictions. Wash. L. Rev., 89:1, 2014

  6. [14]

    The algorithm game

    Jane Bambauer and Tal Zarsky. The algorithm game. Notre Dame L. Rev., 94:1, 2018

  7. [15]

    Working with machines: The impact of algorithmic and data-driven management on human workers

    Min Kyung Lee, Daniel Kusbit, Evan Metsky, and Laura Dabbish. Working with machines: The impact of algorithmic and data-driven management on human workers. In Proceedings of the 33rd annual ACM conference on human factors in computing systems, pages 1603–1612, 2015

  8. [16]

    Are for-profit hospitals really different? medicare upcoding and market structure, 2001

    Elaine Silverman and Jonathan S Skinner. Are for-profit hospitals really different? medicare upcoding and market structure, 2001

  9. [17]

    Efficient algorithms for mining outliers from large data sets

    Sridhar Ramaswamy, Rajeev Rastogi, and Kyuseok Shim. Efficient algorithms for mining outliers from large data sets. InProceedings of the 2000 ACM SIGMOD international conference on Management of data, pages 427–438, 2000

  10. [18]

    Tax fraud detection for under-reporting declarations using an unsupervised machine learning approach

    Daniel De Roux, Boris Perez, Andrés Moreno, Maria del Pilar Villamil, and César Figueroa. Tax fraud detection for under-reporting declarations using an unsupervised machine learning approach. In Proceedings of the 24th ACM SIGKDD International Conference on Knowledge Discovery...

  11. [19]

    Isolation forest

    Fei Tony Liu, Kai Ming Ting, and Zhi-Hua Zhou. Isolation forest. In 2008 eighth ieee international conference on data mining, pages 413–422. IEEE, 2008

  12. [20]

    Extended isolation forest

    Sahand Hariri, Matias Carrasco Kind, and Robert J Brunner. Extended isolation forest. IEEE transactions on knowledge and data engineering, 33(4):1479–1489, 2019. 11

  13. [21]

    Insurance fraud detection with unsupervised deep learning

    Chamal Gomes, Zhuo Jin, and Hailiang Yang. Insurance fraud detection with unsupervised deep learning. Journal of Risk and Insurance, 88(3):591–624, 2021

  14. [22]

    Ecod: Unsu- pervised outlier detection using empirical cumulative distribution functions

    Zheng Li, Yue Zhao, Xiyang Hu, Nicola Botta, Cezar Ionescu, and George H Chen. Ecod: Unsu- pervised outlier detection using empirical cumulative distribution functions. IEEE Transactions on Knowledge and Data Engineering, 35(12):12181–12193, 2022

  15. [23]

    Deep isolation forest for anomaly detection

    Hongzuo Xu, Guansong Pang, Yijie Wang, and Yongjun Wang. Deep isolation forest for anomaly detection. IEEE Transactions on Knowledge and Data Engineering, 2023

  16. [24]

    Learning strategy-aware linear classifiers

    Yiling Chen, Yang Liu, and Chara Podimata. Learning strategy-aware linear classifiers. Ad- vances in Neural Information Processing Systems, 33:15265–15276, 2020

  17. [25]

    Strategic classification made practical

    Sagi Levanon and Nir Rosenfeld. Strategic classification made practical. In International Conference on Machine Learning, pages 6243–6253. PMLR, 2021

  18. [26]

    The social cost of strategic classification

    Smitha Milli, John Miller, Anca D Dragan, and Moritz Hardt. The social cost of strategic classification. In Proceedings of the Conference on Fairness, Accountability, and Transparency, pages 230–239, 2019

  19. [27]

    Incentive-aware pac learning

    Hanrui Zhang and Vincent Conitzer. Incentive-aware pac learning. Proceedings of the AAAI Conference on Artificial Intelligence, 35(6):5797–5804, 2021

  20. [28]

    Learning losses for strategic classification

    Tosca Lechner and Ruth Urner. Learning losses for strategic classification. Proceedings of the AAAI Conference on Artificial Intelligence, 36(7):7337–7344, 2022

  21. [29]

    Strategic classification with unknown user manipulations

    Tosca Lechner, Ruth Urner, and Shai Ben-David. Strategic classification with unknown user manipulations. In International Conference on Machine Learning, pages 18714–18732. PMLR, 2023

  22. [30]

    Manipulation-proof machine learning

    Daniel Björkegren, Joshua E Blumenstock, and Samsun Knight. Manipulation-proof machine learning. arXiv preprint arXiv:2004.03865, 2020

  23. [31]

    Pac-learning for strategic classifica- tion

    Ravi Sundaram, Anil Vullikanti, Haifeng Xu, and Fan Yao. Pac-learning for strategic classifica- tion. Journal of Machine Learning Research, 24(192):1–38, 2023

  24. [32]

    Strategic classification under unknown personal- ized manipulation

    Han Shao, Avrim Blum, and Omar Montasser. Strategic classification under unknown personal- ized manipulation. Advances in Neural Information Processing Systems, 36, 2024

  25. [33]

    Strategic classification from revealed preferences

    Jinshuo Dong, Aaron Roth, Zachary Schutzman, Bo Waggoner, and Zhiwei Steven Wu. Strategic classification from revealed preferences. In Proceedings of the 2018 ACM Conference on Economics and Computation, pages 55–70, 2018

  26. [34]

    Gaming helps! learning from strategic interactions in natural dynamics

    Yahav Bechavod, Katrina Ligett, Steven Wu, and Juba Ziani. Gaming helps! learning from strategic interactions in natural dynamics. In International Conference on Artificial Intelligence and Statistics, pages 1234–1242. PMLR, 2021

  27. [35]

    Statistics and causal inference

    Paul W Holland. Statistics and causal inference. Journal of the American statistical Association, 81(396):945–960, 1986

  28. [36]

    Causality

    Judea Pearl. Causality. Cambridge university press, 2009

  29. [37]

    Causal inference in statistics, social, and biomedical sciences

    Guido W Imbens and Donald B Rubin. Causal inference in statistics, social, and biomedical sciences. Cambridge university press, 2015

  30. [38]

    Nonparametric estimation of heterogeneous treat- ment effects: From theory to learning algorithms

    Alicia Curth and Mihaela Van der Schaar. Nonparametric estimation of heterogeneous treat- ment effects: From theory to learning algorithms. In International Conference on Artificial Intelligence and Statistics, pages 1810–1818. PMLR, 2021

  31. [39]

    Quasi-oracle estimation of heterogeneous treatment effects

    Xinkun Nie and Stefan Wager. Quasi-oracle estimation of heterogeneous treatment effects. Biometrika, 108(2):299–319, 2021

  32. [40]

    https://www.medicare.gov/ what-medicare-covers/your-medicare-coverage-choices

    Your Medicare coverage choices — medicare.gov. https://www.medicare.gov/ what-medicare-covers/your-medicare-coverage-choices . [Accessed 25-04-2024]. 12

  33. [41]

    National neighborhood data archive (nanda): Health care services by census tractstates, 2003-2017 [data set]

    Anam Khan, Mao Li, Jessica Finlay, Michael Esposito, Iris Gomez-Lopez, Philippa Clarke, and Megan Chenoweth. National neighborhood data archive (nanda): Health care services by census tractstates, 2003-2017 [data set]. Inter-University Consortium for Political and Social Resea...

  34. [42]

    Provider of services files, Sep 2022

  35. [43]

    The central role of the propensity score in observational studies for causal effects

    Paul R Rosenbaum and Donald B Rubin. The central role of the propensity score in observational studies for causal effects. Biometrika, 70(1):41–55, 1983

  36. [44]

    Metalearners for estimating heterogeneous treatment effects using machine learning

    Sören R Künzel, Jasjeet S Sekhon, Peter J Bickel, and Bin Yu. Metalearners for estimating heterogeneous treatment effects using machine learning. Proceedings of the national academy of sciences, 116(10):4156–4165, 2019

  37. [45]

    Adapting neural networks for the estimation of treatment effects

    Claudia Shi, David Blei, and Victor Veitch. Adapting neural networks for the estimation of treatment effects. Advances in neural information processing systems, 32, 2019

  38. [46]

    Permutation weighting

    David Arbour, Drew Dimmery, and Arjun Sondhi. Permutation weighting. In International Conference on Machine Learning, pages 331–341. PMLR, 2021

  39. [47]

    Causal effect inference for structured treatments

    Jean Kaddour, Yuchen Zhu, Qi Liu, Matt J Kusner, and Ricardo Silva. Causal effect inference for structured treatments. Advances in Neural Information Processing Systems, 34:24841–24854, 2021

  40. [48]

    Medicare fraud detection using machine learning methods

    Richard A Bauder and Taghi M Khoshgoftaar. Medicare fraud detection using machine learning methods. In 2017 16th IEEE international conference on machine learning and applications (ICMLA), pages 858–865. IEEE, 2017

  41. [49]

    Data reduction techniques for highly imbalanced medicare big data

    John T Hancock, Huanjing Wang, Taghi M Khoshgoftaar, and Qianxin Liang. Data reduction techniques for highly imbalanced medicare big data. Journal of Big Data, 11(1):8, 2024

  42. [50]

    A comparison of state-of- the-art classification techniques for expert automobile insurance claim fraud detection

    Stijn Viaene, Richard A Derrig, Bart Baesens, and Guido Dedene. A comparison of state-of- the-art classification techniques for expert automobile insurance claim fraud detection. Journal of Risk and Insurance, 69(3):373–421, 2002

  43. [51]

    Overpayment models for medical audits: multiple scenarios

    Tahir Ekin, R Muzaffer Musal, and Lawrence V Fulton. Overpayment models for medical audits: multiple scenarios. Journal of Applied Statistics, 42(11):2391–2405, 2015

  44. [52]

    Claims auditing in automobile insurance: fraud detection and deterrence objectives

    Sharon Tennyson and Pau Salsas-Forn. Claims auditing in automobile insurance: fraud detection and deterrence objectives. Journal of Risk and Insurance, 69(3):289–308, 2002

  45. [53]

    Optimal auditing with scoring: Theory and application to insurance fraud

    Georges Dionne, Florence Giuliano, and Pierre Picard. Optimal auditing with scoring: Theory and application to insurance fraud. Management Science, 55(1):58–70, 2009

  46. [54]

    Insurers pocketed $50 billion from medicare for diseases no doctor treated

    Christopher Weaver, Tom McGinty, Anna Wilde Matthews, and Mark Maremont. Insurers pocketed $50 billion from medicare for diseases no doctor treated. Wall Street Journal, 2024

  47. [55]

    Double/debiased machine learning for treatment and structural parameters, 2018

    Victor Chernozhukov, Denis Chetverikov, Mert Demirer, Esther Duflo, Christian Hansen, Whitney Newey, and James Robins. Double/debiased machine learning for treatment and structural parameters, 2018

  48. [56]

    Pyod: A python toolbox for scalable outlier detection

    Yue Zhao, Zain Nasrullah, and Zheng Li. Pyod: A python toolbox for scalable outlier detection. Journal of Machine Learning Research, 20(96):1–7, 2019

  49. [57]

    Pytorch: An imperative style, high-performance deep learning library, 2019

    Adam Paszke, Sam Gross, Francisco Massa, Adam Lerer, James Bradbury, Gregory Chanan, Trevor Killeen, Zeming Lin, Natalia Gimelshein, Luca Antiga, Alban Desmaison, Andreas Köpf, Edward Yang, Zach DeVito, Martin Raison, Alykhan Tejani, Sasank Chilamkurthy, Benoit Steiner, Lu Fan...

  50. [58]

    Fan, Daniel Nouri, Benjamin Bossan, and skorch Developers

    Marian Tietz, Thomas J. Fan, Daniel Nouri, Benjamin Bossan, and skorch Developers. skorch: A scikit-learn compatible neural network library that wraps PyTorch, July 2017. 13

  51. [59]

    Pedregosa, G

    F. Pedregosa, G. Varoquaux, A. Gramfort, V . Michel, B. Thirion, O. Grisel, M. Blondel, P. Prettenhofer, R. Weiss, V . Dubourg, J. Vanderplas, A. Passos, D. Cournapeau, M. Brucher, M. Perrot, and E. Duchesnay. Scikit-learn: Machine learning in Python. Journal of Machine Learni...

  52. [60]

    Oliphant, Matt Haberland, Tyler Reddy, David Cournapeau, Evgeni Burovski, Pearu Peterson, Warren Weckesser, Jonathan Bright, Stéfan J

    Pauli Virtanen, Ralf Gommers, Travis E. Oliphant, Matt Haberland, Tyler Reddy, David Cournapeau, Evgeni Burovski, Pearu Peterson, Warren Weckesser, Jonathan Bright, Stéfan J. van der Walt, Matthew Brett, Joshua Wilson, K. Jarrod Millman, Nikolay Mayorov, Andrew R. J. Nelson, E...

  53. [61]

    CVXPY: A Python-embedded modeling language for convex optimization

    Steven Diamond and Stephen Boyd. CVXPY: A Python-embedded modeling language for convex optimization. Journal of Machine Learning Research, 17(83):1–5, 2016

  54. [62]

    Pfetsch, Franziska Schlösser, Felipe Serrano, Yuji Shinano, Mark Turner, Stefan Vigerske, Dieter Weninger, and Lixing Xu

    Suresh Bolusani, Mathieu Besançon, Ksenia Bestuzheva, Antonia Chmiela, João Dionísio, Tim Donkiewicz, Jasper van Doornmalen, Leon Eifler, Mohammed Ghannam, Ambros Gleixner, Christoph Graczyk, Katrin Halbig, Ivo Hedtke, Alexander Hoen, Christopher Hojny, Rolf van der Hulst, Dom...

  55. [63]

    Harris, K

    Charles R. Harris, K. Jarrod Millman, Stéfan J. van der Walt, Ralf Gommers, Pauli Virtanen, David Cournapeau, Eric Wieser, Julian Taylor, Sebastian Berg, Nathaniel J. Smith, Robert Kern, Matti Picus, Stephan Hoyer, Marten H. van Kerkwijk, Matthew Brett, Allan Haldane, Jaime Fe...

  56. [64]

    successes

    The pandas development team. pandas-dev/pandas: Pandas, 2020. A Additional related works Algorithmic anomaly/fraud detection. Our framework can be understood as an algorithmic audit- ing method for fraud/anomaly detection. Many approaches assume that ground-truth fraud/gaming ...

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.