MEMORANDUM OPINION AND ORDER

Elaine E. Bucklo, United States District Judge.

Relator's second amended complaint in this years-long qui tam action alleges that defendants — three skilled nursing facilities (SNFs) and a therapy provider whose therapists treated patients in each of them — violated the False Claims Act by submitting fraudulent claims to the federal government for reimbursement of services provided to Medicare beneficiaries. Relator's theory is that defendants intentionally ignored the so-called "medical necessity" requirement necessary for reimbursement of skilled nursing services and systematically submitted claims for therapies that: a) were not actually provided to Medicare beneficiaries; and/or b) were provided to Medicare beneficiaries to satisfy defendants' financial objective of maximizing Medicare reimbursements without regard for the beneficiaries' legitimate medical needs. Before me are defendants' motions to exclude the opinions and testimony of relator's experts, Dr. Vivek Shah and Dr. Richard Baer. For the reasons that follow, both motions are granted in part.

I.

The admissibility of expert testimony is governed by Federal Rule of Evidence 702 and the principles established in Daubert v. Merrell Dow Pharm., Inc., 509 U.S. 579, 113 S.Ct. 2786, 125 L.Ed.2d 469 (1993). Under Daubert, trial courts act as gatekeepers to ensure that "any and all scientific testimony or evidence admitted is not only relevant, but reliable." Id. at 589, 113 S.Ct. 2786. The court's inquiry "is fact-dependent and flexible," Lapsley v. Xtek, Inc., 689 F.3d 802, 810 (7th Cir. 2012), and its purpose "is to scrutinize proposed expert witness testimony to determine if it has `the same level of intellectual rigor that characterizes the practice of an expert in the relevant field' so as to be deemed reliable enough to present to a jury." Id. at 805 (quoting Kumho Tire, 526 U.S. 137, 152, 119 S.Ct. 1167, 143 L.Ed.2d 238 (1999)). While the application of the Daubert standard thus depends on the facts of each case, the court must ensure in all cases that the expert is qualified in the relevant field; that the expert's methodology is scientifically reliable; and that the expert's testimony will "assist the trier of fact to understand the evidence or to determine a fact in issue." Fed. R. Evid. 702; Bielskis v. Louisville Ladder, Inc., 663 F.3d 887, 893-94 (7th Cir. 2011). The court must not, however, supplant the jury's "ageless role" of "evaluating witness credibility and weight of the evidence." Stollings v. Ryobi Technologies, Inc., 725 F.3d 753, 766 (7th Cir. 2013). Provided an expert employs a methodology recognized in his or her field of expertise and formulates an opinion based on "the kinds of facts or data" on which others in the field would reasonably rely, vigorous cross-examination and the presentation of contrary evidence are generally the appropriate tools for exposing flaws in the expert's analysis or conclusions. Manpower, Inc. v. Insurance Co. of Pennsylvania, 732 F.3d 796, 809 (7th Cir. 2013). See also Daubert, 509 U.S. at 596, 113 S.Ct. 2786.

Understanding where relator's proposed expert testimony fits into his theory of liability requires a brief explanation of how the Centers for Medicare and Medicaid Services ("CMS") reimburses SNFs for the services they provide to beneficiaries. Part A of the Medicare program, at issue here, provides federally-funded insurance benefits for, among other things, skilled nursing and rehabilitation care. Medicare pays SNFs using a prospective payment system ("PPS"). PPS payments are per diem, per patient rates based primarily on a patient's expected therapeutic needs, which are classified as falling within one of several Resource Utilization Groups ("RUGs"). A clinical assessment tool known as the Minimum Data Set ("MDS") is performed periodically to determine each beneficiary's RUG level as of each Assessment Reference Date ("ARD") and its corresponding look-back period. See 42 C.F.R. §§ 413.337, 413.343; Medicare Program Integrity Manual 6.1.1 There are five therapy-related RUG levels, which range from "Ultra High" for beneficiaries receiving at least 720 minutes of therapy per week in at least two disciplines (among occupational therapy, physical therapy, and speech language pathology) with at least one of those disciplines treating the beneficiary five days per week, to "Low" for beneficiaries who receive forty-five minutes per week of total therapy. See 63 Fed. Reg. at 26,262. "Very High," "High," and "Medium" RUG designations fall within these two poles.

Relator alleges that defendants systematically "upcoded" Medicare patients, meaning that they provided them with excessive therapies for the purpose of satisfying the criteria for the highest possible RUG classifications, regardless of whether the therapies were necessary or beneficial to the patients based on their clinical conditions. Defendants achieved this, relator claims, by "ramping" services during the assessment reference periods, i.e., providing inflated therapy during these periods (which would determine reimbursement rates prospectively until the next assessment period), then decreasing therapy intensity during non-assessment periods.

Dr. Shah, whom relator proffers as an expert in economics and data analysis, examined data provided by defendants (the "Casamba" data in his first report, to which defendants refer as "Shah I," and CMS Cost Report data in his second report, "Shah II," and opines based on these data that: 1) therapists working at defendants' facilities recorded more time providing therapy to patients than the time they were actually present at the facilities; and 2) defendants systematically provided more intensive services during the RUG assessment periods than they provided during non-RUG assessment periods, and that this systematic discrepancy cannot be explained by medical necessity or other patient-specific factors. Dr. Shah also extrapolates from the data he analyzed the estimated damages in the form of Medicare overpayments that the government incurred as a result of the alleged scheme.

Dr. Baer is a physician and former Medical Director for the Medicare contractors who administer claims under Medicare Part A (relevant here as it applies to inpatient skilled nursing services).2 Dr. Baer's expert report summarizes the legislative and regulatory requirements for the payment of Medicare claims, focusing specifically on the requirement that services for which reimbursement is sought must be "reasonable and necessary for the diagnosis or treatment of illness or injury or to improve the functioning of a malformed body member," and must meet other criteria established by CMS. Dr. Baer opines that defendants' policies, including SimplyRehab's "Standards for Therapy Department Performance," are indicative of a pattern of fraudulent activity designed to inflate Medicare reimbursements, and that Dr. Shah's data confirm that the implementation of these policies "subverted" the medical necessity determinations required for claims payment.

II.

Defendants articulate numerous arguments for excluding the testimony of both experts in toto. While I agree that methodological flaws warrant the exclusion of certain of each expert's opinions, I am satisfied that portions of their respective reports satisfy Daubert's three-prong standard.

A. Dr. Shah

Relator states that Dr. Shah "performed statistical analysis to evaluate whether there is evidence consistent with an exaggeration of therapy minutes right before Medicare's assessment reference dates (`ARD')." Defendants first argue that Dr. Shah is not qualified to render an opinion about "classifications under the RUG system," but this argument is a red herring. Dr. Shah does not purport to opine about the appropriateness of any specific RUG classification observed in the data he examined; instead, he performed a multifactor regression analysis of the kind statisticians and microeconomists undisputedly employ to arrive at his opinion that defendants systematically inflated RUG levels during assessment periods. Nothing in defendants' submissions persuades me that only an expert in Medicare administration can perform scientifically valid statistical analysis of data relating to the provision of services to Medicare beneficiaries. Defendants vastly overread Klaczak v. Consolidated Medical Transport, 458 F.Supp.2d 622 (N.D. Ill 2006), for this argument. There, the court concluded that a paramedic and lawyer whose experience included reviewing "some contracts with hospitals" for compliance with Medicare billing practices, attending American Ambulance Association conferences, and going on one ride with the Chicago fire department, and who had "no formal or other training in economics or other quantitative discipline" was not qualified to perform a "quantitative analysis of the market conditions in the ambulance-services industry." Id. at 666. The problem in Klaczak was a mismatch between the expert's area of expertise and the nature of the opinion he offered. Here, there is no such mismatch: Dr. Shah is a microeconomist and statistician who performed statistical analysis of data produced in this case. I am satisfied that he is qualified to render the opinions he offers.

Defendants also attack the reliability of Dr. Shah's opinions on multiple grounds. First, they argue that his methodology has not been used, tested, or peer-reviewed. That is, while defendants do not dispute that multiple regression analysis of the kind Dr. Shah performed is a widely accepted statistical tool, they argue that it has not previously been employed or studied as a means of detecting Medicare overpayments. The purpose of Dr. Shah's regression analysis, however, is not to identify specific Medicare overpayments, but rather to examine and quantify the extent to which the aggregate level of therapy defendants provided to Medicare beneficiaries in a given period depended on whether the period under consideration was a RUG assessment period or a non-RUG assessment period. That Dr. Shah cannot point to another expert who has performed a regression analysis to examine the relationship between these specific variables does not negate the reliability of his model.

Nor does defendants' insistence that Dr. Shah used an "unverified" or "wrong" data set warrant exclusion. Defendants observe that neither the Casamba data nor the CMS Cost Reports reflect the actual claims they submitted to CMS for reimbursement. But those data unquestionably have a "qualitative and quantitative connection" to Dr. Shah's analysis. Manpower, Inc., v. Ins. Co. of Pennsylvania, 732 F.3d 796, 808 (7th Cir. 2013). In Manpower, the Seventh Circuit explained that while the data must have a rational connection to the expert's analysis, "arguments about how the selection of data inputs affects the merits of the conclusions produced by an accepted methodology should normally be left to the jury." Id. (citing cases).

Defendants also complain that Dr. Shah fails to consider "obvious alternative explanations" for the abrupt drop he perceives in therapy intensity at the end of Medicare beneficiaries' RUG-assessment periods. Defendants offer examples drawn from a handful of patient charts that appear to reflect episodes of patient illness or the occurrence of holidays falling on non-RUG-assessment days. But even a layperson can grasp the flimsiness of this argument: defendants offer no reason to think that illnesses or holidays are statistically more likely to occur on non-RUG-assessment days than on RUG-assessment days, so this "alternative explanation" is no answer at all to the global trend Dr. Shah's model reveals. Defendants also criticize Dr. Shah for failing to appreciate that patients are expected to improve over time such that therapy intensity should likewise be expected to decrease over time. But according to Dr. Shah, the data tell a different story. See Shah II at ¶¶ 35-35 and Fig. 6. Whether Dr. Shah's analysis can withstand the contrary interpretation of defendant's rebuttal expert, Dr. Salve, is a matter the jury can determine.

Neither Brown v. Burlington Northern Santa Fe Ry. Co., 765 F.3d 765 (7th Cir. 2014), nor Barron v. Deloitte & Touche, LLP, 2008 WL 7136869 (W.D. Tex. 2008), persuades me that Dr. Shah's report should be stricken for failure to consider obvious alternative explanations for his findings. Brown involved an injured railroad worker who sued his employer for the "cumulative trauma disorder" he claimed was caused by his working conditions, but whose medical expert failed to investigate whether other activities he knew the plaintiff engaged in — notably, motorcycle-riding and volunteer firefighting — could have caused the injuries. Id. at 770. Nothing about the court's exclusion of the expert testimony in that case disturbs my view that defendant's cherry-picked examples of patient illness or other episodic causes of missed therapy sessions do not call into question the reliability of Dr. Shah's statistical findings. And in Barron, the relator's expert developed a valid random sample of 1600 claims; determined that 991 of those claims could be tested; and opined that 911 of them were "invalid" because they were not supported by sufficient documentation. The expert explicitly based his conclusion on "an assumption that all documents that were available to support the sample items should have been retained by [the defendants] and produced to Relator's Counsel," notwithstanding affirmative evidence that at least one defendant had not retained the documents, and despite making no independent effort to determine whether the documents could be found elsewhere. Because the expert's opinion that each of the 911 individual claims was "invalid" based on a common, untested, and apparently counterfactual assumption that all supporting documentation that had ever existed had been made available to him, the court deemed his opinion unreliable. But as noted above, Dr. Shah does not purport to pass upon the "validity" of any specific claim. Relator's ultimate ability to prove, by some means, that one or more of the actual claims defendants submitted to CMS were fraudulent will be put to the test at summary judgment or at trial, but the question presently before me is not whether Dr. Shah's analysis is sufficient to establish liability under the FCA. As I understand it, Dr. Shah purports to show that defendants systematically provided more therapy to Medicare patients on RUG-assessment days than they provided on non-RUG-assessment days even controlling for other factors that might influence therapy length or intensity. Indeed, the summary of his opinion on this issue is that: "[t]otal therapy length during `assessment periods' which are used to determine the rate of reimbursement by Medicare was significantly higher than during other non-assessment periods."3 This opinion may not get relator across the finish line, but it is grounded in the kind of statistical analysis that statisticians routinely employ.

Where Dr. Shah runs into trouble, however, is in his effort to estimate damages based on trends in the data that do not account for individual patient conditions. A central assumption in Dr. Shah's damages calculations is that the level of therapy Medicare beneficiaries received during non-RUG assessment periods represents their "true therapy needs." See Shah I at ¶ 45. While that assumption is consistent with relator's theory that defendants "inflated" the amount of therapy provided during RUG-assessment days in order to increase the rate at which they were reimbursed, it finds no basis in the facts Dr. Shah considered, nor is it obvious as a matter of reason or common sense. After all, it does not follow from relator's theory that defendants abandoned individualized medical necessity determinations in favor of default classifications driven by defendants' bottom-line that none of the default classifications was, in fact, appropriate to the patient's needs. Presumably, at least some of defendants' patients genuinely needed the highest levels of therapy, even if defendants failed to undertake the patient-specific medical necessity determinations to reach that conclusion prior to assigning them to the highest RUG classification.

Of course, that does not explain the discrepancy in therapy intensity between assessment and non-assessment periods. But even assuming that the trend Dr. Shah observes in the data raises a plausible inference that something is rotten in the state of Denmark, one could easily infer a very different sort of scheme than the one relator asserts. Indeed, assuming that the "true therapy needs" of at least some of defendants' Medicare patients corresponded to the high RUG levels they were assigned during RUG-assessment periods, one might argue based on Dr. Shah's data that defendants routinely cut corners by failing to provide adequate therapy to those patients during non-assessment periods, while still receiving reimbursement payments at the higher RUG levels. Arguably, either scheme would violate the FCA, but the damages calculation would be very different, as illustrated below.

Because Dr. Shah's model assumes that the lower RUG levels are the "correct" ones, he considers all patient days billed at the higher levels to be fraudulent, regardless of the amount of therapy actually provided on those days. He then proposes two, alternative, damages calculations: one that assumes full disgorgement of every "fraudulent" patient day, i.e., the entire amount the SNF received for each day coded at the higher level; and a second that assumes partial disgorgement of each "fraudulent" patient day, i.e., the margin between the amount the SNFs actually received and the amount they would have received if they had correctly coded the patients at the lower RUG level. Under a "cutting corners" theory, by contrast, "fraudulent" days would presumably be those on which patients whose conditions warranted the level of therapy commensurate with the higher RUG classifications, but who did not, in fact, receive high levels of therapy. Unlike in Dr. Shah's model, the damages produced by a "cutting corners" theory are not agnostic to the level of therapy defendants actually provided.

What the foregoing brings into focus is that even if Dr. Shah's statistical analysis is reliable evidence of defendants' practice of systematically decreasing the amount of therapy provided to Medicare beneficiaries at the end of their RUG-assessment periods, damages, if any, cannot reliably be calculated, or even estimated, in the absence of any information about any Medicare beneficiary's genuine medical needs. Indeed, the entire premise of relator's theory is that legitimate RUG classifications must be based on individualized medical necessity determinations. Not only does Dr. Shah's damages model fail to account for medical necessity, it is premised on the unsupported and implausible yet plainly material assumption that none of the comparatively higher RUG classifications reflect the patient's true therapy needs. For at least these reasons, his damages estimates do not clear Daubert's reliability hurdle.

B. Dr. Baer

The first two pages of Dr. Baer's eleven-page expert report recount his credentials as an expert in Medicare administration and the review of potentially fraudulent claims. Baer Rep., Def.'s Mem., Exh. 1 at 1-2. Defendants do not dispute Dr. Baer's expertise in this area. The next three-and-a-half pages are devoted to a description of the Prospective Payment System, and in particular, of the steps therapy service providers must follow to establish the "medical necessity" of skilled nursing services provided to Medicare beneficiaries. Id. at 3-6. Defendants also do not challenge this portion of his report.

The contested portion of Dr. Baer's report begins on page 6, with the section captioned, "SimplyRehab's and the Skilled Nursing Facilities' Policies Subverted Medical Necessity Determinations by Licensed Professionals." In this section, Dr. Baer reproduces portions of two articles written by Robert Kunio (identified in one of the publications as SimplyRehab's Chief Operating Officer, though Dr. Baer does not identify Kunio's relationship to the parties, and relator refers to him in his opposition memorandum as SimplyRehab's president) and other documents setting forth SimplyRehab's "Standards for Therapy." Focusing first on Kunio's description of "a `benchmarking' methodology to increase the reimbursement for therapy services in the SNF setting," Dr. Baer quotes excerpts in which Kunio states that the methodology accomplished "a 200% increase in the percentage of Med-A days that are Ultra High for the long-term care population in the nursing homes we serve" and "a 20% increase in average [length of stay] before discharge from therapy." Id. at 6-7. Dr. Baer observes that the article omits any mention of the medical necessity requirement or the need for licensed professionals to determine the intensity and duration of therapy and opines that "many times ... when business goals drive the rendering of medical services, rather than the clinical decisions of healthcare professionals, that's when providers conduct abusive and/or fraudulent practices, by circumventing the medical necessity safeguards provided by unbiased professional clinical judgments." Id.

Dr. Baer goes on to opine that SimplyRehab's therapy standards "turned the determination of what therapy a patient receives on its head, with reimbursement levels driving the provision of services, rather than medical necessity." Id. at 8. Dr. Baer points to Kunio's description of the role of "Rehab Managers" as "encourag[ing] staff to always think twice and err on the side of more treatment and longer lengths of stay." Id. In Dr. Baer's opinion, Kunio's statement is further evidence that defendants' "management continually instituted policies and procedures that usurped the independent professional judgment of its licensed therapists in order to meet unrealistic business goals." Id. Finally, Dr. Baer opines that Dr. Shah's report:

Id. at 8.

Dr. Baer then offers four "patient examples" that he states are "representative of the pervasive pattern of activity in patient care that provided excessive minutes of therapy for excessive periods of time, and treated patients who did not require a skilled nursing facility level of care." Id. at 9-10. He summarizes each patient's condition at the time of admission, and any evaluations or notes in the file and concludes that in each case, the therapy plan of care was excessive, not medically necessary, and inappropriate for the patient's condition.

Dr. Baer concludes his report with the opinion that "SimplyRehab developed and implemented a scheme indicative of a pattern of fraudulent activity to inappropriately increase Medicare Parts A and B reimbursement." Defendants achieved this, in Dr. Baer's opinion, by:

Id. at 11.

As noted above, defendants acknowledge Dr. Baer's expertise in the field of Medicare administration and the review of Medicare claims. The thrust of their objection is that his opinions are not derived from the facts he would have considered or the analyses he would have undertaken in the various medical director roles in which he developed his expertise.

Defendants' forty-five page Daubert motion blasts Dr. Baer as unqualified to "vouch" for Dr. Shah's analysis; derides his opinions as "derived from his fraud-detecting and mind-reading skills" that are neither reliable nor helpful to the jury; and insists that Dr. Baer's summary review of a tiny, non-random sample of patient charts cannot be extrapolated to the general population of defendants' patients. The first of these arguments is overblown: Defendants assert that Dr. Baer, lacking training in statistics, should not be allowed to "accept[], champion[], and rel[y] on Dr. Shah's work" to show a pattern of misconduct. But Dr. Baer does not "champion" Dr. Shah's findings; he relies on them in the ordinary way experts in one field commonly rely on the opinions of experts in another field. See Dura Automotive Systems of Indiana, Inc. v. CTS Corp., 285 F.3d 609, 613 (7th Cir. 2002) ("[I]t is common in technical fields for an expert to base an opinion in part on what a different expert believes on the basis of expert knowledge not possessed by the first expert....").4 The remainder of defendants' challenge to Dr. Baer's reliance on Dr. Shah essentially regurgitates the arguments I rejected above as a basis for excluding Dr. Shah's testimony.

Defendants second and third arguments, however, have merit. Where Dr. Baer's opinions falter is not in their reliance on Dr. Shah's statistical findings but in the leap his opinions take from those findings, Mr. Kunio's statements about "benchmarking," documents describing SimplyRehab's therapy standards, and a cursory review of four patient files to the inference of a collective scheme to commit widespread fraud. It bears emphasis that unlike Dr. Shah's opinions, which, as noted above, stop short of asserting that the patterns he detects in the data are evidence of fraud, Dr. Baer opines affirmatively that defendants "developed and implemented a scheme indicative of a pattern of fraudulent activity" to inflate Medicare reimbursements without regard for medical necessity.

Conspicuously, the methodology Dr. Baer used to discern fraud in this case is decidedly different from the patient-specific one he insists is required by Medicare rules and regulations. Indeed Dr. Baer's emphasis on the individualized nature of the medical necessity inquiry undercuts the statistics-based methodology he applied in this case. Worse, even assuming that a statistics-based methodology can be used to prove FCA fraud — an assumption for which U.S. v. Rogan, 517 F.3d 449, 453 (7th Cir. 2008), offers some support, although U.S. ex rel. Crews v. NCS Healthcare of Illinois, Inc., 460 F.3d 853, 857 (7th Cir. 2006), arguably points in the other direction, and courts around the country are divided, see U.S. ex rel. Michaels v. Agape Senior Community, Inc., C/A No. 0:12-3466-JFA, 2015 WL 3903675, at *7 (D.S.C. 2015) (disallowing statistical sampling to prove liability or damages under the FCA and citing cases reflecting split of authority) — Dr. Baer admitted that there was "no statistical methodology" to the selection of his probe sample of four patients, Baer Dep., Def.'s Mot. at Exh. 2 at 52:5-6. Yet, in a recent case in which Dr. Baer testified as an expert, he acknowledged that a statistically valid sample is necessary to extrapolate observations drawn from a limited sample group to a larger patient population. Baer Dep. in U.S. v. Paulus, 15-CR-15 (E.D. Ky.), Def.'s Mot. at Exh. 3 at 15-99-15-100.

Further debilitating Dr. Baer's analysis is his acknowledgement that he did not undertake the CMS-prescribed claim review process even with respect to the probe sample of patient charts he reviewed. Indeed, despite professing his conviction that defendants engaged in "the plainest and clearest fraudulent scheme [he has] ever come across," Baer Dep. at 69:9-10, and that the examples he selected are "illustrative," id. at 50:3, he admits that he did not "perform a review that met the standards that [he] used when [he was] doing claim reviews as a medical director," id. at 50:20-51:1 and cannot determine on the basis of the information he reviewed that any particular claim defendants submitted was fraudulent, see id. at 61:2-14 (explaining that "a more granular" review would be required in which he would "look at the claims, look at the billing, see what was billed on any particular patient," among other things).

These shortcomings in Dr. Baer's analysis infect his overarching opinion that defendants "developed and implemented a scheme indicative of a pattern of fraudulent activity" designed to inflate Medicare reimbursements. This is not a problem that can be left to sort out at summary judgment. Unlike Dr. Shah's opinions (other than on damages), which are the product of a widely recognized methodology that statisticians commonly employ to detect trends in data irrespective of whether those trends are sufficient to establish any element of relator's claim, Dr. Baer's opinion goes directly to the question of fraud. Yet his opinion that defendants engaged in fraud is disassociated from any of the CMS-prescribed methodologies he discusses in his report and testimony. In fact, Dr. Baer admits that his opinions are based not on any methodology he would have employed as a medical director, but on his "judgment" of admittedly partial evidence. Id. at 67:8. His analysis, in short, was both less rigorous and different in kind than any he would have conducted in his regular professional work. Daubert requires more. See Gopalratnam v. Hewlett-Packard Co., 877 F.3d 771, 780 (7th Cir. 2017); Lapsley v. Xtek, Inc., 689 F.3d 802, 805 (7th Cir. 2012).

As defendants correctly observe, there is no such thing as a "fraud expert." Fraud requires an inference about a defendant's mental state, and it is counsel's job — not an expert witness's — to establish a link, through argument between the evidence and the defendant's intent. See In re Rezulin Products Liability Litigation, 309 F.Supp.2d 531, 545 (S.D.N.Y. 2004). In this case, relator is entitled to present evidence, through Dr. Baer, of CMS's claims review process and the criteria that must be met to establish medical necessity. And while I decline, at this point, to exclude categorically Dr. Bear's testimony about the kinds of fraudulent schemes he and his colleagues uncovered in the course of their professional duties, he may not testify to his abstract judgment that defendants' conduct fits the bill. More specific determinations about the scope of Dr. Baer's proposed testimony are unnecessary at this juncture.

III.

For the foregoing reasons, defendants' motions to exclude the testimony of relator's proposed experts are granted in part as set forth above.