Wood, Chief Judge.
*654Anyone who has encountered medical billing, either from the provider side or the patient side, would agree that it could kindly be described as complex. But that does not mean that the rules are impossible to decipher, or that anything goes. There is a line between honest billing and fraud, and in the present case, the government charged Dr. Omeed Memar with crossing it.
The government focused on eight cases in which Memar diagnosed patients with a precancerous condition known as actinic keratosis, which produces distinctive skin lesions. He treated the patients in question by applying intense pulsed light ("IPL") to supposed lesions on their faces and then billing their insurance providers for the procedures. The government suspected that there were no such lesions, and that instead this was a ruse to trick insurance companies into reimbursing the charges for cosmetic procedures outside the scope of the policies. A jury agreed with the government. On appeal, Memar primarily insists that the evidence was insufficient to support the verdict. We are not persuaded: we see ample evidence in this record from which the jury could find that Memar's conduct was fraudulent. We find no reversible error elsewhere either, and so the convictions must stand.
I
Until recently, Memar was a practicing dermatologist in Chicago. In 2015, he was indicted on eight counts of health care fraud, in violation of 18 U.S.C. § 1347, and eight counts of making false statements affecting a health care matter, in violation of 18 U.S.C. § 1035(a)(2). These charges relate to Memar's care for the eight patients we described above. After each session of IPL treatment for the alleged actinic keratosis, Memar submitted claims to the patients' insurance companies, using billing code 17004, for the destruction of 15 or more precancerous lesions. The insurance companies paid, since the paperwork represented that the treatments were medically necessary and not cosmetic.
At trial, the government presented two theories of fraud: it contended that the eight patients named in the indictment did not actually have actinic keratosis ; and it asserted that the use of IPL alone could not destroy the lesions produced by actinic keratosis. In support of the first theory, the prosecution presented expert testimony from Dr. Edward Victor Ross, who explained that actinic keratosis most commonly presents as red scaly lesions and is uncommon among patients under 40. Dr. Ross also opined that IPL alone, without other substances or devices, was ineffective at treating actinic keratosis.
The eight patients, who were mostly in their 20s and 30s, all took the stand. Many of them stated that Memar never told them that they had actinic keratosis or a precancerous condition, something they said they would have remembered. They had not seen scaly spots on their skin. To their knowledge, the IPL was to address other medical problems or to improve their skin's appearance. Another dermatologist, *655Dr. Neha Robinson, confirmed that two of the patients did not have lesions when she saw them during the same period as the one in which they were receiving IPL treatments from Memar.
Memar's medical assistants also testified. They reported that at each IPL visit, they copied into the patients' charts diagrams of where each actinic keratosis lesion was located. They did this without examining the patients' faces at the time of the visit. Two of the assistants said that they would write down that a patient had multiple lesions even if they observed none. One assistant admitted that she believed she "was doing treatments on patients that didn't have these diagnoses."
The government additionally offered testimony from a representative of Blue Cross Blue Shield (BCBS), James Krupkowski, about a meeting he and others had with Memar. BCBS suspected that Memar used IPL on a patient for cosmetic reasons, and so Krupkowski confronted him about it. Memar responded, according to Krupkowski, that the IPL was for "prevention," since the patient "had a proven history of cancer." The prosecution introduced records showing that shortly after this meeting, Memar dramatically reduced the frequency with which he billed for the destruction of 15 or more actinic keratosis lesions. Patients testified that they were told their IPL treatments would no longer be covered, but they were not offered any alternative treatments-surprising if the treatment was for precancerous conditions.
In his defense, Memar called eight other patients with actinic keratosis. These people, some of whom had had biopsies to confirm their disease, were generally older and knew they had precancerous lesions. They testified that Memar discussed multiple treatment options with them, and they underwent other procedures in addition to IPL. The defense also called an expert, Dr. David Goldberg. Dr. Goldberg did not employ IPL alone to treat actinic keratosis in his practice, but he did say that he used it to treat his wife's actinic keratosis at home.
Finally, Memar testified in his own defense. He stood by his diagnoses and maintained that the efficacy of unsupplemented IPL, though investigational, had some support in the field. On cross-examination, Memar repeatedly denied the prosecution's suggestion that he admitted in the meeting with BCBS that he used IPL for photorejuvenation (a cosmetic purpose).
Though Memar and Krupkowski's testimony was clear that there had been no such admission, the prosecution falsely stated three times during its closing argument that Memar told BCBS that he used IPL for photorejuvenation. Memar's attorney did not object, but he did not let the comments slide. In his closing statement, defense counsel quickly turned to this mischaracterization of testimony. He reminded the jury that neither Krupkowski nor Memar testified to an admission about photorejuvenation. A few minutes later, the defense attorney used the misstatement to attack the government's credibility, citing the alleged "confession" to BCBS as proof that the jury was "not getting the truth." But this was not the end of the matter. In its rebuttal, the prosecution referred again to the same nonexistent testimony while listing evidence against Memar. The defense did not object.
Ultimately, the jury convicted Memar on all 16 counts. The court declined Memar's motion under Federal Rules of Criminal Procedure 29 and 33 to enter a judgment of acquittal or to grant a new trial. He was sentenced to a three-year period of supervised release, during which he must perform community service and pay fines and restitution. On appeal, Memar repeats his two reasons for attacking his convictions:
*656that the evidence was insufficient to prove all elements of the offenses, and that the prosecution's improper remarks tainted the verdict.
II
Free access — add to your briefcase to read the full text and ask questions with AI
Wood, Chief Judge.
*654Anyone who has encountered medical billing, either from the provider side or the patient side, would agree that it could kindly be described as complex. But that does not mean that the rules are impossible to decipher, or that anything goes. There is a line between honest billing and fraud, and in the present case, the government charged Dr. Omeed Memar with crossing it.
The government focused on eight cases in which Memar diagnosed patients with a precancerous condition known as actinic keratosis, which produces distinctive skin lesions. He treated the patients in question by applying intense pulsed light ("IPL") to supposed lesions on their faces and then billing their insurance providers for the procedures. The government suspected that there were no such lesions, and that instead this was a ruse to trick insurance companies into reimbursing the charges for cosmetic procedures outside the scope of the policies. A jury agreed with the government. On appeal, Memar primarily insists that the evidence was insufficient to support the verdict. We are not persuaded: we see ample evidence in this record from which the jury could find that Memar's conduct was fraudulent. We find no reversible error elsewhere either, and so the convictions must stand.
I
Until recently, Memar was a practicing dermatologist in Chicago. In 2015, he was indicted on eight counts of health care fraud, in violation of 18 U.S.C. § 1347, and eight counts of making false statements affecting a health care matter, in violation of 18 U.S.C. § 1035(a)(2). These charges relate to Memar's care for the eight patients we described above. After each session of IPL treatment for the alleged actinic keratosis, Memar submitted claims to the patients' insurance companies, using billing code 17004, for the destruction of 15 or more precancerous lesions. The insurance companies paid, since the paperwork represented that the treatments were medically necessary and not cosmetic.
At trial, the government presented two theories of fraud: it contended that the eight patients named in the indictment did not actually have actinic keratosis ; and it asserted that the use of IPL alone could not destroy the lesions produced by actinic keratosis. In support of the first theory, the prosecution presented expert testimony from Dr. Edward Victor Ross, who explained that actinic keratosis most commonly presents as red scaly lesions and is uncommon among patients under 40. Dr. Ross also opined that IPL alone, without other substances or devices, was ineffective at treating actinic keratosis.
The eight patients, who were mostly in their 20s and 30s, all took the stand. Many of them stated that Memar never told them that they had actinic keratosis or a precancerous condition, something they said they would have remembered. They had not seen scaly spots on their skin. To their knowledge, the IPL was to address other medical problems or to improve their skin's appearance. Another dermatologist, *655Dr. Neha Robinson, confirmed that two of the patients did not have lesions when she saw them during the same period as the one in which they were receiving IPL treatments from Memar.
Memar's medical assistants also testified. They reported that at each IPL visit, they copied into the patients' charts diagrams of where each actinic keratosis lesion was located. They did this without examining the patients' faces at the time of the visit. Two of the assistants said that they would write down that a patient had multiple lesions even if they observed none. One assistant admitted that she believed she "was doing treatments on patients that didn't have these diagnoses."
The government additionally offered testimony from a representative of Blue Cross Blue Shield (BCBS), James Krupkowski, about a meeting he and others had with Memar. BCBS suspected that Memar used IPL on a patient for cosmetic reasons, and so Krupkowski confronted him about it. Memar responded, according to Krupkowski, that the IPL was for "prevention," since the patient "had a proven history of cancer." The prosecution introduced records showing that shortly after this meeting, Memar dramatically reduced the frequency with which he billed for the destruction of 15 or more actinic keratosis lesions. Patients testified that they were told their IPL treatments would no longer be covered, but they were not offered any alternative treatments-surprising if the treatment was for precancerous conditions.
In his defense, Memar called eight other patients with actinic keratosis. These people, some of whom had had biopsies to confirm their disease, were generally older and knew they had precancerous lesions. They testified that Memar discussed multiple treatment options with them, and they underwent other procedures in addition to IPL. The defense also called an expert, Dr. David Goldberg. Dr. Goldberg did not employ IPL alone to treat actinic keratosis in his practice, but he did say that he used it to treat his wife's actinic keratosis at home.
Finally, Memar testified in his own defense. He stood by his diagnoses and maintained that the efficacy of unsupplemented IPL, though investigational, had some support in the field. On cross-examination, Memar repeatedly denied the prosecution's suggestion that he admitted in the meeting with BCBS that he used IPL for photorejuvenation (a cosmetic purpose).
Though Memar and Krupkowski's testimony was clear that there had been no such admission, the prosecution falsely stated three times during its closing argument that Memar told BCBS that he used IPL for photorejuvenation. Memar's attorney did not object, but he did not let the comments slide. In his closing statement, defense counsel quickly turned to this mischaracterization of testimony. He reminded the jury that neither Krupkowski nor Memar testified to an admission about photorejuvenation. A few minutes later, the defense attorney used the misstatement to attack the government's credibility, citing the alleged "confession" to BCBS as proof that the jury was "not getting the truth." But this was not the end of the matter. In its rebuttal, the prosecution referred again to the same nonexistent testimony while listing evidence against Memar. The defense did not object.
Ultimately, the jury convicted Memar on all 16 counts. The court declined Memar's motion under Federal Rules of Criminal Procedure 29 and 33 to enter a judgment of acquittal or to grant a new trial. He was sentenced to a three-year period of supervised release, during which he must perform community service and pay fines and restitution. On appeal, Memar repeats his two reasons for attacking his convictions:
*656that the evidence was insufficient to prove all elements of the offenses, and that the prosecution's improper remarks tainted the verdict.
II
Memar first argues that there was insufficient evidence to prove the three elements required for his convictions: the falsity of his representations, the willfulness of his false statements, and his intent to defraud. See United States v. Chhibber , 741 F.3d 852, 858-59 (7th Cir. 2014) (defining a section 1347 offense); United States v. Natale , 719 F.3d 719, 742 (7th Cir. 2013) (same for section 1035(a)(2) ).
In evaluating Memar's challenge to the sufficiency of the evidence, we ask not whether we would have found Memar guilty in the first instance. Rather, we consider "only if, after viewing the evidence in the light most favorable to the government, the record is devoid of evidence from which a rational trier of fact could find guilt beyond a reasonable doubt." Chhibber , 741 F.3d at 858. This deferential inquiry, articulated by the Supreme Court in Jackson v. Virginia , 443 U.S. 307, 99 S.Ct. 2781, 61 L.Ed.2d 560 (1979), in the context of collateral review but widely accepted as a general standard, prevents us from reweighing evidence or assessing witness credibility. United States v. Moshiri , 858 F.3d 1077, 1081-82 (7th Cir. 2017). All we can do is look at the record as a whole and ask whether the evidence would permit a reasonable jury to find each element. United States v. Farmer , 717 F.3d 559, 563 (7th Cir. 2013). Memar may not simply relitigate his defense on direct appeal. Cf. Natale , 719 F.3d at 743.
Moreover, contrary to Memar's assertions, there is nothing wrong with circumstantial evidence of guilt. United States v. Jones , 713 F.3d 336, 340 (7th Cir. 2013) ; United States v. Galati , 230 F.3d 254, 258 (7th Cir. 2000). While a verdict based on speculation cannot stand, one premised on reasonable inferences is sound. Jones , 713 F.3d at 340 ; see also United States v. Daniel , 749 F.3d 608, 614 (7th Cir. 2014) (circumstantial evidence and inferences from the scheme to defraud can establish specific intent to defraud); United States v. Rucker , 738 F.3d 878, 884 (7th Cir. 2013) (circumstantial evidence can be used to establish knowing participation in a scheme to defraud).
A
We look first at the question whether the evidence supported the jury's finding that Memar purposefully misdiagnosed patients as having actinic keratosis. The government presented substantial proof to demonstrate the falsity of those diagnoses. Numerous patients testified that they had zero-let alone over 15-scaly patches on their faces. Dr. Robinson confirmed that two of the patients did not have actinic keratosis. One of Memar's medical assistants admitted that she wrote in patient charts that actinic keratosis lesions were present when they in fact were not. Another said that she did not notice any scaly plaques, either at the initial visit or at the time of treatment. Many of the patients in question were in their 30s or younger, a population in which actinic keratosis is possible but "abnormal." Dr. Ross testified that 95% of the time, young patients with actinic keratosis have it only on their lips. In his over 20 years of practice, Dr. Ross has never had a patient in this age range exhibit 15 or more lesions on other parts of their faces.
Memar insists that the absence of visibly scaly patches does not rule out the possibility of actinic keratosis, since the condition can manifest in various ways. There *657was testimony that actinic keratosis could take the form of brown spots or developing lesions underneath the skin. But Memar's charts belie this defense. The records describe the patients' actinic keratosis as "scaly" plaques, macules, and patches-the classic symptoms. The fact that they could have had atypical or preclinical actinic keratosis is thus irrelevant.
Viewing this evidence in the light most favorable to the prosecution, there was more than enough proof for a jury to conclude that the eight patients did not in fact have actinic keratosis. See Chhibber , 741 F.3d at 860 (evidence that the defendant documented unreported medical symptoms and diagnoses supported convictions under sections 1347 and 1035 ).
The record also contained evidence suggesting that Memar lied about the presence of lesions with the intent to defraud the insurance companies. After BCBS confronted Memar about his use of IPL in what it suspected was a cosmetic procedure, he started requiring patients to pay for IPL out-of-pocket. Memar's preferred inference-that he stopped billing patients' insurance for IPL because the insurer would no longer cover it-may have been possible from the facts, but the jury was not required to accept it, particularly given Memar's failure to offer the affected patients alternative actinic keratosis treatments.
And why would these patients have expected other treatments, when they did not know they had actinic keratosis ? The contrasting care received by the patients named in the indictment and those Memar called on his behalf corroborates the theory that Memar willfully made false diagnoses. The latter group was generally older and familiar with their actinic keratosis. Indeed, Memar told an FBI agent that he always informed patients with actinic keratosis when they had it. Yet the mostly younger patients named in the indictment were not informed, either by Memar or anyone in his office, that they had actinic keratosis or a precancerous condition. The bona fideactinic keratosis patients discussed their medical options with Memar and received treatments in addition to IPL. The patients in the indictment did not. Instead, they often thought they were receiving IPL for cosmetic or other medical reasons.
Further evidence of Memar's intent comes from the medical assistants' testimony. Memar instructed them, they said, to note in the charts the number and placement of facial lesions at each IPL treatment. But they were to do so by copying what was recorded at the initial visit-not by examining the patient's face at the time of the later appointment. A jury could infer willfulness from this practice. See Chhibber , 741 F.3d at 860 (evidence that defendant ordered tests on patients whom he did not examine and who did not report the recorded symptoms supported section 1347 counts); Natale , 719 F.3d at 743 (record was not " 'devoid' of evidence on intent" where jury could infer from defendant's detailed notes and other evidence that chart inaccuracies were the result of "knowing deliberation rather than careless inadvertence").
On this record, the jury could infer that in order to make insurance companies pay for something that otherwise fell outside their policies, Memar performed a cosmetic procedure on patients whom he intentionally misdiagnosed. The evidence against him is neither speculative nor a reflection of a difference in medical opinion. Memar presented his defense at trial, and "[t]he jury disbelieved that story." Id . We will not disturb its findings. See United States v. Kolbusz , 837 F.3d 811, 812 (7th Cir. 2016).
*658B
We could stop here and affirm Memar's convictions on the government's first theory of liability. We think it useful, however, to explain why the evidence also supports the government's second theory: that Memar committed fraud by submitting claims that IPL by itself (that is, unsupplemented by other measures) destroyed 15 or more lesions, something he knew it could not do.
IPL, when not used in conjunction with an agent, is not a standard treatment for actinic keratosis. Nonetheless, Memar testified that he sincerely believed in IPL's potential. In addition, Dr. Goldberg opined that IPL can destroy lesions and testified about experiments finding IPL somewhat effective. Dr. Ross disputed the accuracy of these studies. According to him, the literature does not support the hypothesis that IPL alone "works or if it works, it works very poorly and very temporarily." The jury was thus presented with conflicting evidence. It was entitled to find more credible Dr. Ross's view that IPL is inadequate by itself.
We recognize that doctors often use medications and treatments in off-label ways. If they have a good-faith belief in the potential success of their experimentation, they do not commit health-care fraud. Some of the trial testimony could have been understood as suggesting that Memar's use of IPL was a permissible off-label approach. For example, Memar mentioned in a 2016 presentation that IPL should be studied further as a potential treatment for actinic keratosis. The jury could have thought that the fact that he considered this method worthy of additional attention implies that he honestly believed in its promise.
Other evidence, however, supported the opposite conclusion. Cf. Kolbusz , 837 F.3d at 812. Memar's office performed IPL treatments without conducting any follow-up examination to determine which patients had any lesions in need of removal, and where any such lesions were located. The jury was entitled to find that the staff were insincere about their efforts to destroy lesions when they did not bother to ascertain where the lesions were. See United States v. Hunt , 521 F.3d 636, 645 (6th Cir. 2008) (upholding a conviction for health care fraud where circumstantial evidence revealed that defendant "caused bills to be submitted ... for patients that he had never seen and tests that had not been determined to be medically necessary"). The jury was also entitled to give weight to the fact that, for patients with biopsy-confirmed actinic keratosis, Memar's behavior was quite different. For the latter group, he made sure to provide other treatments in addition to IPL. In sum, the evidence supported a jury finding that Memar intended to defraud the insurance companies by claiming he was successfully performing a medically necessary procedure.
III
Memar's other argument on appeal is that the prosecution's mischaracterization of key testimony during closing statements so infected his trial with unfairness that his conviction amounts to a violation of due process. See Darden v. Wainwright , 477 U.S. 168, 181, 106 S.Ct. 2464, 91 L.Ed.2d 144 (1986). The government concedes that its repeated misstatements were inappropriate. Thus, the only question before us is whether the improper remarks deprived Memar of a fair trial. United States v. Klemis , 859 F.3d 436, 442 (7th Cir. 2017).
Our answer to this question turns on five factors: "(1) the nature and seriousness of the alleged misconduct; (2) whether the defense invited the prosecutor's statements; (3) whether the jury *659instructions adequately addressed the matter; (4) whether the defense had an opportunity to respond to the improper remarks; and (5) the weight of the evidence against the defendant." Id . Because Memar did not object in the district court to the prosecutor's comments, we review this argument under the familiar plain-error standard. See Puckett v. United States, 556 U.S. 129, 135, 129 S.Ct. 1423, 173 L.Ed.2d 266 (2009). As relevant here, that means that even if our analysis of the prosecutorial misconduct uncovers error that normally would lead to reversal, Memar can prevail only if the error is obvious, it affected his substantial rights, and it seriously affects the "fairness, integrity, or public reputation of judicial proceedings." Id.
Turning first to the analysis of the prosecutorial misconduct, we find that the first two criteria favor Memar. "Misstatement of evidence may be improper prosecutorial conduct." United States v. Wolfe , 701 F.3d 1206, 1214 (7th Cir. 2012). And the prosecutor's remarks were unforced error. But the rest of our analysis convinces us that the trial, viewed in its entirety, was fair.
The district court instructed the jury before and after closing arguments that the lawyers' statements were not evidence. "If what a lawyer said is different from the evidence as you remember it," he told the jury, "the evidence is what counts." We have suggested that such general instructions can sometimes be enough to ensure a trial's fairness where the mischaracterization does not lie at the heart of the case. See Jordan v. Hepp , 831 F.3d 837, 849 (7th Cir. 2016) ; United States v. McMath , 559 F.3d 657, 668 (7th Cir. 2009). Although the parties quibble over whether this "admission" is essential for the government's second theory of criminal liability, neither considers it essential to the first.
To his credit, Memar's defense attorney took advantage of his closing argument to combat the prosecution's misrepresentations. See United States v. Bowman , 353 F.3d 546, 552 (7th Cir. 2003). He reminded the jury of the actual testimony and was able to use that testimony to discredit the prosecution. The fact that the government got in one more jab during rebuttal, to which defense counsel did not respond, is insufficient to render the entire trial unfair. See McMath , 559 F.3d at 668 (trial's fairness was not affected even though the improper comment was made during the government's rebuttal).
Critically, there was plenty of evidence against Memar apart from the testimony about what happened at the BCBS meeting. We frown upon the government's repeated errors, but we cannot say that they tainted the fairness of the entire trial.
IV
In the same vein, Memar argues that his trial attorney's failure to object to the prosecutor's remarks amounted to ineffective assistance of counsel under Strickland v. Washington , 466 U.S. 668, 104 S.Ct. 2052, 80 L.Ed.2d 674 (1984).
We have strongly cautioned against raising ineffective assistance claims on direct appeal. E.g. , Ramirez v. United States , 799 F.3d 845, 853 (7th Cir. 2015) ; United States v. Flores , 739 F.3d 337, 340-41 (7th Cir. 2014). To prevail on such a claim, Memar would have to present evidence to overcome the strong presumption that his attorney was engaged in reasonable trial strategy. United States v. Jansen , 884 F.3d 649, 656 (7th Cir. 2018) ; Hicks v. Hepp , 871 F.3d 513, 525 (7th Cir. 2017). "[A] decision not to object to clearly objectionable comments in the prosecution's closing argument" can be at times "sound trial strategy."
*660United States v. Driver , 798 F.2d 248, 255 (7th Cir. 1986). He has not done so at this time. Without the benefit of an evidentiary hearing or a developed record, we have no idea why Memar's counsel decided to wait until his closing to say something. See Vinyard v. United States , 804 F.3d 1218, 1227 (7th Cir. 2015) (ineffective assistance claims "generally depend on information outside the record available on direct appeal").
This does not leave Memar without a remedy. If he files a post-conviction motion under 28 U.S.C. § 2255, he may seek an opportunity to develop the needed record. The custody requirement for such a motion will not be a problem for some time, as Memar will remain in custody until his supervised release ends in 2020. See, e.g. , Clarke v. United States , 703 F.3d 1098, 1101 (7th Cir. 2013) (supervised release is a form of custody that renders collateral relief available); Kusay v. United States , 62 F.3d 192, 193 (7th Cir. 1995) (same). If we were to consider the merits of Memar's Sixth Amendment argument now, he would be barred from pursuing it in a later proceeding. Flores , 739 F.3d at 341-42. Given that Memar did not stress this claim at oral argument, we think it best to refrain from reaching it now. See Vinyard , 804 F.3d at 1227 (recognizing that we have allowed the withdrawal of "claims raised on direct appeal when oral argument made clear that those claims would benefit from additional factual development"); United States v. Wallace , 753 F.3d 671, 676 (7th Cir. 2014) (on direct appeal dismissing without prejudice the defendant's Strickland claims lest they be forfeited).
V
Memar's trial was far from "devoid" of evidence that he acted fraudulently. A jury was entitled to find that Memar engaged in the charged misconduct. And while the prosecution's remarks were improper, they did not deprive Memar of a fair trial. Accordingly, we AFFIRM the judgment of the district court.