Sandra Blakes sought Supplemental Security Income disability benefits for her minor son, Lamanuel Wolfe, Jr. An Administrative Law Judge found that Laman-uel was not disabled, the Appeals Council denied his request for review, and the district court affirmed the ALJ’s ruling, finding it was supported by substantial evidence. Because the ALJ failed to apply the proper standard in considering Laman-uel’s claim, we vacate and remand.
I.
Blakes filed an application for Supplemental Security Income (“SSI”) disability benefits with the Social Security Administration (“SSA”) on behalf of her son. La-manuel, born August 25, 1993, was five years old as of the effective date of his application and six years old at the time of his hearing before the ALJ. Blakes presented evidence to the ALJ that Lamanuel was being treated for a seizure disorder and also received services for speech and language delays. Various medical records indicated that Lamanuel suffered from an “atypical seizure disorder,” a “seizure disorder,” and a “stable seizure disorder.” R. 12, at 105, 116, 139. Lamanuel took two medications to control his disorder but nonetheless occasionally still suffered seizures. According to Milwaukee Public School records, Lamanuel underwent testing in February 1999 that showed he was delayed 17 months to three years in speech intelligibility and receptive and expressive language. R. 12, at 85. In intelligence testing, Lamanuel received scores of 70 in visual reasoning and short-term memory, 97 in verbal reasoning, 98 in quantitative reasoning, and 81 composite. R. 12, at 81. On a Peabody Picture Vocabulary Test, Lamanuel scored a 40, which *567 placed him below the first percentile. 1 R.12, at 80. Lamanuel repeated kindergarten at the recommendation of his teacher. R. 12, at 62,129.
An examiner at the Wisconsin Hearing Aid Centers evaluated Lamanuel in February 1999 and found that he exhibited delays in following sequential instructions and receptive vocabulary. He also lacked a good vocabulary for some basic items. The examiner stated that Lamanuel had a “relatively good prognosis” for achieving age-appropriate language functioning, and suggested that some of the delays may have been caused by environmental factors. However, the examiner also emphasized that “[e]xtremely important at this point in time is to rule out any learning problems via psychometrics.” R. 12, at 112-13.
The ALJ accepted as valid the intelligence testing scores. She acknowledged that Lamanuel was receiving services for speech and language delays and that he was being treated for a possible seizure disorder. She noted the speech and language assessment which demonstrated severe delays in speech intelligibility and receptive and expressive language. She also remarked on Lamanuel’s lack of social or behavioral problems and noted that his kindergarten teacher attributed some of his delays to immaturity. The ALJ rejected Lamanuel’s claim that he met the requirements of the listing for mental retardation based on several IQ subscores of 70:
The requirements of section 112.05 are satisfied if an individual has at least one score of 60 to 70 and another severe impairment. His speech/language delays could be considered “another severe impairment.” However, the examiner in [exhibit] 3F specifically stated that the claimant has good prognosis and that his language problems caused only minimal effects on his activities of daily living. This is not a description of a person who is mentally retarded. His attorney argued that his mother is retarded and receives benefits for the condition. This appears to be relevant in suggesting that the claimant’s delays are caused by poor environmental stimulation and can be remedied by an enriched school environment.
See Decision, R. 12, at 14. The ALJ found on the basis of the evidence that Lamanuel suffered from severe impairments, namely speech and language delays and a possible seizure disorder. However, she also found that his condition did not meet or medically equal the requirements of any impairment listed in Appendix 1, Subpart P, Regulations No. 404. The ALJ ruled that Lamanuel’s impairments did not cause him to functionally equal the requirements of any section of the listing, that he suffered no disabling limitations from a chronic illness or from his treatment and medications, and that he has no disabling limitations in broad areas of functioning. She found marked impairment in communication and cognition based on “substantial current delays in speech and language and over-all low average intellectual functioning.” R. 12, at 14. But she also found there was no evidence that the delays are permanent. Id. She found less than marked impairment in motor functioning due to a possible mild seizure disorder, less than marked impairment in personal functioning and in concentration, and no impairment in social functioning. R. 12, at 15. On that basis, she concluded that, although Lamanuel was suffering from two severe impairments (speech and language *568 delays and a seizure disorder), he was not disabled as that term is defined in the Social Security Act. Id.
The SSA Appeals Council denied La-manuel’s request for review of the ALJ’s decision, and Lamanuel then appealed to the district court. R. 12, at 4-5. A magistrate judge recommended that the ALJ’s denial of benefits be affirmed because her conclusion that Lamanuel is not mentally retarded was supported by substantial evidence. The district court adopted the report and recommendation of the magistrate judge, finding that the ALJ adequately supported her conclusion that Lamanuel was not mentally retarded. See Blakes v. Barnhart, No. 01-CV-7 (E.D.Wisc. March 7, 2002). The district court also found that the magistrate considered all of the evidence, even the evidence favorable to Lamanuel in reaching this conclusion and that because of the district court’s limited review, the denial of benefits must be affirmed. Lamanuel appeals.
II.
On appeal, Lamanuel contends that his impairment meets the requirements of fisting 112.05(D), that the ALJ failed to build a logical bridge between the evidence and her conclusions, that the ALJ relied on her own judgment about the cause of Lamanuel’s impairments without any medical support in the record for that judgment, that the ALJ should have called upon a medical expert under the circumstances, and that Lamanuel’s seizure disorder and speech and language impairments meet the requirements of the second prong of fisting 112.05(D). Lamanuel also complains that the ALJ failed to determine the credibility of key witnesses and failed to discuss significant evidence favoring his claim.
Because the SSA’s Appeals Council denied Lamanuel’s request for review, the ALJ’s ruling constitutes the final decision of the Commissioner.
See Scott v. Barnhart,
With those standards in mind, we turn to the provision on which Lamanuel rests his claim for disability:
112.05 Mental Retardation: Characterized by significantly subaverage general intellectual functioning with deficits in adaptive functioning.
The required level of severity for this disorder is met when the requirements in A, B, C, D, E, or F are satisfied....
D. A valid verbal, performance, or full scale IQ of 60 through 70 and a physical or other mental impairment imposing an additional and significant limitation of function[.]
20 C.F.R. Part 404, Subpart P, Appendix 1, Rule 112.05. As we noted above, the ALJ found that Lamanuel had valid IQ *569 scores of 70 in both visual reasoning and short-term memory. She also found that he suffered from other severe impairments, namely speech and language delays and a possible seizure disorder. She nonetheless concluded that Lamanuel did not meet the listing because he is not mentally retarded. Because the speech examiner predicted a good prognosis and stated that Lamanuel’s speech and language problems caused only minimal effects on his activities of daily living, the ALJ found that Lamanuel could not meet the listing. She also opined that Lamanuel’s delays could have been caused by poor environmental stimulation because his mother has been diagnosed as mentally retarded, and she suggested that Lamanuel’s problems could therefore be remedied by an enriched school environment. She thus concluded that Lamanuel’s “condition does not meet or medically equal the requirements of any impairment listed in Appendix 1, Subpart P, Regulations No. 404” and that his impairments do not cause him to “functionally equal the requirements of any section of the listings.” R. 12, at 14.
We require the ALJ to build an accurate and logical bridge from the evidence to her conclusions so that we may afford the claimant meaningful review of the SSA’s ultimate findings.
Scott,
Here, the ALJ failed to build that bridge, instead relying on conjecture and her own assessment of the medical evidence to reach conclusions unsupported by the record. For example, the ALJ opined that a person with a severe speech and language disorder who nonetheless has a good prognosis cannot be considered mentally retarded. In drawing this conclusion, the ALJ apparently assumed that a child who could improve his language skills was not retarded. We say “apparently” because we see no medical evidence in the record tying together Lamanuel’s speech and language problems with his low IQ scores, for example. Indeed, the speech examiner emphasized that it was extremely important to rule out any learning problems via psychometrics. The ALJ’s conclusion that a possibility for improvement in speech equals a lack of mental retardation has no more support in the record than a conclusion that an improvement in Lamanuel’s seizure disorder would equal a lack of mental retardation. There is simply no basis in the record to tie the two disorders together and the ALJ should not have assumed a connection in the absence of record evidence.
See Steele,
*570
The ALJ also concluded that Lamanuel’s mental retardation may have been the result of poor environmental stimulation because of his mother’s diagnosis of the same disability, and that his problems could thus be remedied by an enriched school environment. Here again, the ALJ appears to have been reaching conclusions unsupported by evidence in the record. There is little evidence in the record regarding the origin of Lamanuel’s low IQ scores and no evidence that an enriched school environment would remedy the situation. This assessment is the result of a hunch and an ALJ may not rely on a hunch.
Wilder v. Chater,
Before we depart, we must comment on a dispute brewing between the parties that will affect the handling of the case on remand. At the time the ALJ rendered a decision in this case, the Code of Federal Regulations stated that
Listing 112.05 (Mental Retardation) contains six sets of criteria, any one of which, if satisfied, will result in a finding that the child’s impairment meets the listing.
20 C.F.R. Pt. 404, Subpt. P, App. 1, § 112.00A (2000). We have previously concluded that under this formulation, the regulations did not require that a claimant meet the diagnostic definition of the listing as well as one of the six sets of criteria.
See Scott,
If an impairment satisfies the diagnostic description of the introductory paragraph and any one of the six sets of criteria, we will find that the child’s impairment meets the listing.
20 C.F.R. Pt. 404, Subpt. P, App. 1, § 112.00A (2001) (emphasis added). It is not apparent from our review of the ALJ’s decision whether she applied the first formulation or the second. We suspect she applied the second because she appears to have found that Lamanuel meets the criteria of one of the six sets (he has an IQ of 70 and another severe impairment), and yet she concluded he did not meet the *571 listing, that he is not mentally retarded. On remand, she should apply the formulation in effect at the time of Lamanuel’s hearing. That is, if he meets any one of the six sets of criteria in the listing, he meets the listing. She should not require that he also show that he satisfies the diagnostic description. of the introductory paragraph.
Vacated AND Remanded.
Notes
. Although the scale used in the Peabody Picture Vocabulary Test differs from the more familiar IQ tests, Lamanuel's test scores were all consistent when translated to percentile rankings.
See Scott v. Barnhart,
