Margaret Hayward appeals the decision of the district court affirming the Department of Labor’s (DOL) decision to deny survivor benefits for the death of her husband. The Energy Occupational Illness Compensation Program Act, 42 U.S.C. § 7384 et seq. (the Act), provides, among other things, for the lump-sum payment to survivors of persons who contracted cancer as a result of exposure to radiation during employment in the United States nuclear weapons program. Survivors may receive benefits under the Act “only if’ there is a 50% or greater probability that the decedent contracted cancer from exposure to radiation during employment in a covered facility. Id. at §§ 73841(9)(B), 7384n(b). To make this determination, the DOL is required to use interactive software that estimates the probability of causation for various types of cancer using certain “default settings.” 42 C.F.R. §§ 81.20, 81.21. The sole issue in this appeal is whether the DOL acted arbitrarily and capriciously in refusing to adjust these default settings in light of the decedent’s exceedingly rare form of prostate cancer. Under the deferential standard of review required, because the DOL set forth a rational connection between the relevant factors and its decision to retain the default settings, the district court did not err in granting summary judgment upholding the DOL’s decision. As a result, we affirm.

I. STATUTORY SCHEME

Congress enacted The Energy Occupational Illness Compensation Program Act to provide benefits to employees with illnesses caused by exposure to radiation and other toxic substances in the course of their work for the Department of Energy (DOE) or its predecessor agencies, and certain of its contractors and subcontractors. See 42 U.S.C. § 7384. Under Part B of the Act, employees or their eligible survivors can receive a lump-sum payment of $150,000 for certain illnesses, including cancer, caused by exposure to radiation in the course of employment at Department of Energy facilities. Id. at §§ 73841(9)(A) and (B), 7384s(a)(l). An employee or his or her survivors are entitled to compensation under Part B of the Act “if, and only if, the cancer ... was at least as likely as not related to employment” in a covered facility. Id. at § 7384n(b).

An individual seeks benefits under Part B of the Act by filing a claim with the DOL’s Office of Workers’ Compensation Programs (OWCP). ^[1] See 20 C.F.R.[*378] §§ 30.100, 30.101. When a claim is made, the OWCP gathers the employee’s relevant factual and medical information and transfers it to the National Institute for Occupational Safety and Health (NIOSH) ^[2] to perform a radiation dose reconstruction. The dose reconstruction estimates the amount of radiation received by the employee during covered employment using a variety of factors including the employee’s age, gender, employment history, workplace characterization data, and any other information useful for characterizing workplace radiation exposure. 42 C.F.R. § 82.14. The OWCP uses the completed dose reconstruction to determine whether the employee’s cancer is “at least as likely as not related to employment” in a covered facility. See 42 U.S.C. § 7384n(b).

The OWCP bases its probability calculation on an interactive-computer-software program specifically designed for adjudication of claims under Part B of the Act, the NIOSH-IREP. 42 C.F.R. § 81.20. The NIOSH-IREP “models the dose-response relationship between ionizing radiation and 33 cancers using morbidity data from the ... Japanese atomic bomb survivor cohort.” Id. at § 81.10(a). The program uses these models coupled with NIOSH’s dose reconstruction to calculate the probability that the employee developed cancer due to radiation exposure during employment in a covered facility. The program’s default settings account for uncertainty from several sources, including statistical uncertainty in the various cancer risk models. Id. at § 81.11. The program, however, contains a feature entitled “User Defined Additional Uncertainty” that “can be adjusted to account for the presence of additional uncertainty and bias correction not presently” accounted for by the program’s default settings. The program’s-manual states that the default settings should only be adjusted “after sufficient justification accompanied by a written rationale.”

After OWCP finishes its probability of causation calculation, it issues a recommended decision on the claim. 20 C.F.R. § 30.305(a). The claimant can file objections within 60 days of the date of the recommended decision. Id. at § 30.310(a). Once adjudication of the claim is complete, the Final Adjudication Branch within the OWCP issues the DOL’s final decision.

II. FACTUAL AND PROCEDURAL BACKGROUND

Margaret Hayward is the surviving widow of Milton Hayward. Mr. Hayward worked in the DOE’s nuclear weapons program for over fifteen years, during which time he was exposed to ionizing radiation. On February 3, 1997, doctors diagnosed Mr. Hayward with sarcomatoid carcinoma, an exceedingly rare form of prostate cancer. ^[3] The cancer spread rapidly and he died shortly thereafter.

Ms. Hayward filed a claim for compensation under Part B of the Act based on her husband’s death. As required by the Act, the OWCP gathered Mr. Hayward’s relevant factual and medical information[*379] and transferred the claim to NIOSH to perform a radiation dose reconstruction. OWCP then used the dose reconstruction and the NIOSH-IREP program to calculate a 21.41% probability that Mr. Hayward developed cancer from radiation exposure during his employment with the DOE. Based on this figure, the OWCP issued a recommended decision denying Ms. Hayward’s claim.

Ms. Hayward objected to the OWCP’s recommended decision and requested a hearing. Although she lodged several objections, she primarily contended that the OWCP significantly underestimated the probability that radiation caused Mr. Hayward’s cancer due to its failure to consider the rarity of sarcomatoid carcinoma and its alleged higher correlation with radiation exposure (as compared to other forms of prostate cancer). At the hearing, Ms. Hayward provided medical literature regarding cancer patients treated with radiation therapy and a letter from Mr. Hayward’s doctor to support these claims. Ms. Hayward contended that this evidence justified adjusting the default settings of the “User Defined Uncertainty Distribution” within the NIOSH-IREP program.

Following the hearing, the DOL hearing representative sought guidance from a DOL physicist concerning the propriety of adjusting the default settings of the NIOSH-IREP program based on Ms. Hayward’s objections. Nearly a month later, and after consulting with a representative of the developer of the program, the DOL physicist concluded that Ms. Hayward’s objections did not justify altering the program’s default settings. Thereafter, the OWCP’s Final Adjudication Branch issued a final decision denying Ms. Hayward’s survivor claim.

Ms. Hayward filed an original complaint in the Northern District of Texas, alleging that the DOL acted arbitrarily and capriciously in refusing to adjust the default settings of the NIOSH-IREP program to account for the rarity of Mr. Hayward’s cancer. Subsequently, the Director of the Division of Energy Employees Occupational Illness Compensation Program, acting under his regulatory authority, vacated the OWCP’s final decision and returned Ms. Hayward’s claim to the Final Adjudication Branch for a revised final decision addressing in more detail Ms. Hayward’s objections. The revised final decision more thoroughly addressed Ms. Hayward’s objections by incorporating the findings of the DOL physicist. Ms. Hayward then filed an amended complaint in the Northern District, again arguing that the DOL acted arbitrarily and capriciously in denying her claim. The DOL moved for summary judgment on this claim, arguing that it was entitled to judgment as a matter of law because the administrative record showed that the OWCP considered the relevant factors and issued a final decision rationally related to those factors. The district court granted summary judgment for the DOL. Ms. Hayward appeals that decision.

III. STANDARD OF REVIEW

This Court reviews the grant of summary judgment de novo, applying the same standard to review the DOL decision as the district court. Templet v. HydroChem Inc., 367 F.3d 473, 477 (5th Cir.2004). Because Part B of the Act does not contain a standard of review and does not require that a formal hearing be held, the district court correctly reviewed the OWCP’s final decision under the arbitrary and capricious standard set forth in section 706(2)(A) of the Administrative Procedures Act.

The arbitrary and capricious standard is “highly deferential,” and we must afford the agency’s decision “a presumption of regularity.” United States v. [*380] Garner, 767 F.2d 104, 116 (5th Cir.1985) (citation omitted). We limit our review to whether the agency “articulated a rational connection between the facts found and the decision made,” Pension Benefit Guar. Corp. v. Wilson N. Jones Mem’l Hosp., 374 F.3d 362, 367 (5th Cir.2004), and “it is well-settled that an agency’s action must be upheld, if at all, on the basis articulated by the agency itself.” Motor Vehicle Mfrs. Ass’n v. State Farm Mut. Auto. Ins. Co., 463 U.S. 29, 42-43, 103 S.Ct. 2856, 77 L.Ed.2d 443 (1983). Although we must conduct a “searching and careful review” of the administrative record to determine whether the agency acted in an arbitrary and capricious manner, we may not substitute our judgment for the agency’s. Citizens to Preserve Overton Park, Inc. v. Volpe, 401 U.S. 402, 415-416, 91 S.Ct. 814, 28 L.Ed.2d 136 (1971). Our mandate is not to “weigh the evidence pro and con but to determine whether the agency decision was based on a consideration of relevant factors and whether there was a clear error of judgment.” Delta Found., Inc. v. United States, 303 F.3d 551, 562 (5th Cir. 2002) (internal quotations omitted). Courts will “uphold a decision of less than ideal clarity if the agency’s path may reasonably be discerned.” Nat’l Ass’n of Home Builders v. Defenders of Wildlife, — U.S. -, 127 S.Ct. 2518, 2530, 168 L.Ed.2d 467 (2007) (quoting Bowman Transp., Inc. v. Arkansas-Best Freight System, Inc., 419 U.S. 281, 286, 95 S.Ct. 438, 42 L.Ed.2d 447 (1974)). In reviewing technical agency decisions such as these, “[w]e must look at the decision not as a chemist, biologist, or statistician that we are qualified neither by training nor experience to be, but as a reviewing court exercising our narrowly defined duty of holding agencies to certain minimal standards of rationality.” Gulf Restoration Network v. United States Dep’t of Transp., 452 F.3d 362, 368 (5th Cir.2006) (citation omitted).

IV. DISCUSSION

This case presents the very circumstance contemplated by Gulf Restoration. Ms. Hayward asks the court to disregard the agency’s expertise and substitute its own judgment of scientific literature unaccompanied by expert analysis. Ms. Hayward’s sole contention is that the OWCP acted arbitrarily and capriciously in refusing to adjust the default settings of the NIOSH-IREP program to account for the rare nature of her deceased husband’s cancer. She bases this argument on two related grounds. First, she argues that sar-comatoid carcinoma, because of its rarity, cannot be fairly compared to the typical adenocarcinoma, the type of prostate cancer most represented in the dose-risk relationship model used by the NIOSH-IREP program to determine the probability of causation for all types of prostate cancer. Second, she argues that sarcomatoid carcinoma should be treated differently than other forms of prostate cancer because, she contends, it is far more likely to be caused by exposure to radiation.

Having reviewed the administrative record, we agree that the OWCP did not engage in arbitrary and capricious decision making when it chose to retain the default settings of the NIOSH-IREP program. The OWCP, after consulting with a DOL physicist, issued a final decision that specifically addressed both of Ms. Hayward’s objections. With respect to Ms. Hayward’s objection concerning the rarity of sarcomatoid carcinoma, the OWCP responded that the excess relative risk (ERR) of prostate (or any) cancer is determined by the following formula: ERR/ dose [Cancer risk attributable to radiation exposure] [divided by] [base-link risk (of that cancer to the general population)[*381] [plus] cancer risk attributable to radiation exposure]. It further stated:

The prostate cancer morphology (i.e., adenocarcinoma or sarcomatoid carcinoma) does not affect the ERR/dose and the resultant NIOSH-IREP result. Any difference (or rarity) in the type of prostate cancer is already accounted for in the definition of ERR/dose; the incidence rate of each of these prostate cancers is inherently reflected within the “general population” cohort. NIOSH-IREP uses the ERR/does data for “All Male Genitalia” from the Japanese Atomic Bomb survivors, and therefore, would be inclusive of all types of prostate cancers.

Thus, the OWCP specifically considered Ms. Hayward’s objection concerning the rarity of sarcomatoid carcinoma but determined that the dose-risk model that the NIOSH-IREP program uses to calculate the probability of causation for all types of prostate cancer already accounts for this factor. In other words, the general prostate cancer data used by the OWCP, while based primarily on adenocarcinoma (because it is more common), includes cases of sarcomatoid carcinoma as well and thus yields an accurate probability of causation for both types of prostate cancer.

Indeed, the practice of grouping rare cancers in more general cancer categories was specifically intended by HHS — the agency charged with the task of developing a method of computing causation under the Act. HHS regulations indicate that the failure to group rare cancers with more common cancers would provide the rare cancers with an unfair advantage due to the high level of uncertainty about their dose-risk relationship. ^[4] See Guidelines for Determining Probability of Causation Under the Employees Energy Occupational Illness Compensation Program Act of 2000; Final Rule, 67 Fed.Reg. 22296, 22302 (May 2, 2002). To counter this uncertainty, HHS noted that the dose-risk models used by NIOSH would group rarer cancers into more general cancer categories. Id. Thus, in refusing to adjust the default settings of the NIOSH-IREP program to account for the rarity of sarcoma-toid carcinoma, the OWCP operated the program consistent with its intended use.

With respect to Ms. Hayward’s argument that the NIOSH-IREP program failed to account for the greater probability that radiation exposure caused her husband’s cancer, the OWCP’s final decision responded:

• Radiation is only one of many potentially cancer-causing agents; the carcinogenic effects of radiation cannot be distinguished from any other toxic agent. Further, it is a generally accepted fact that prostate cancer is less radiogenic than many other cancers.

• There is no evidence that sarcomatoid carcinoma of the prostate is highly correlated with radiation exposure, nor that this rare form of prostate cancer (sarcomatoid) is more radiogenic than the more general form (adenocarci-noma)

In its initial decision, the OWCP more fully explained its statement that there is “no evidence” that sarcomatoid carcinoma is more radiogenic than adenocarcinoma. The OWCP noted that the:

literature [Ms. Hayward] submitted on the correlation on this point refers to radiation that was received as part of a therapeutic treatment for the [adenocar-cinoma] form of prostate cancer. Such[*382] treatment involves a much more focused and localized administration of radiation than would occur in an occupational setting.

The literature submitted by Ms. Hayward suggests that individuals already suffering from the adenocarcinoma form of prostate cancer are more likely to develop sarcomatoid carcinoma if they undergo radiation therapy as treatment. The OWCP specifically noted, and Ms. Hayward does not contest, that Mr. Hayward was diagnosed with only one primary cancer, sarco-matoid carcinoma. Thus, the OWCP could rationally conclude that this literature is inapplicable to Mr. Hayward’s situation. ^[5]

Under our limited scope of review,' the question is not whether sarcomatoid carcinoma is more radiogenic than other forms of prostate cancer, whether Mr. Hayward contracted cancer from radiation exposure during his employment, or even whether the OWCP used the best method to make this determination. Rather, the question is a narrow one: Whether the OWCP considered the relevant factors and provided a reasoned basis for its decision to deny Ms. Hayward’s claim? After reviewing the administrative record, we conclude that the OWCP satisfied this minimum threshold and thus its denial of Ms. Hayward’s survivor claim was not arbitrary or capricious.

Ms. Hayward also contends that the lack of specific standards for adjusting the User Defined Additional Uncertainty feature of the NIOSH renders any decision of the DOL arbitrary and capricious. Given Ms. Hayward’s failure to establish that Mr. Hayward’s form of cancer warrants an adjustment, this argument is unavailing. See Sierra Club v. U.S. Fish and Wildlife Service, 245 F.3d 434, 446 (5th Cir.2001) (deferring to agency to resolve ambiguities). Agencies are not required to proceed by set standards in order to avoid a finding that their actions are arbitrary and capricious. See SEC v. Chenery Corp., 332 U.S. 194, 203, 67 S.Ct. 1575, 91 L.Ed. 1995 (1947) (noting that agencies “must retain power to deal with [ ] problems on a case-by-case basis if the administrative process is to be effective.”). It is sufficient that the DOL considered the relevant factors and provided a reasoned basis for its decision to retain the default settings in this case.

Finally, Ms. Hayward asserts that the Act must be liberally construed to effectuate its remedial purpose of compensating workers that developed diseases while serving in the nation’s nuclear weapons program. Although we do not disagree with Ms. Hayward’s characterization of the Act, its remedial purpose cannot expand the narrow scope of our review under the Administrative Procedures Act. See Norfolk S. Railway Co. v. Sorrell, 549 U.S. 158, 127 S.Ct. 799, 808, 166 L.Ed.2d 638 (2007) (noting that a “statute’s remedial purpose cannot compensate for the lack of statutory basis.”).

The administrative record in this case plainly reveals a rational connection between the relevant facts and the DOL’s decision to deny Ms. Hayward’s Part B claim. Our limited review ends there. Accordingly, we affirm the district court’s summary judgment order.

AFFIRMED.