Table 2.
Popularity and accessibility of electronic health record (EHR) question answering (QA) datasets. We have listed the number of citations and the number of studies on EHR QA using the dataset. The information presented here is based on the data available on September 30, 2023.
| Datasets | Number of citations | Number of studies on EHR QA using the datasets | Publicly available |
| emrQA [26] | 151 | 11 | Yes |
| MIMICSQL [5] | 51 | 3 | Yes |
| Yue et al [46] | 40 | 0 | No |
| MIMICSPARQL*a [41] | 27 | 2 | Yes |
| Yue et al [42] | 18 | 0 | Yes |
| Roberts and Demner-Fushman [23] | s18 | 3 | No |
| emrKBQA [8] | 15 | 0 | No |
| Raghavan et al [34] | 13 | 0 | No |
| Roberts and Demner-Fushman [24] | 10 | 1 | No |
| Soni et al [44] | 7 | 3 | No |
| Fan [35] | 7 | 0 | Yes |
| DrugEHRQA [25] | 5 | 0 | Yes |
| DiSCQb [43] | 6 | 0 | Yes |
| Oliveira et al [38] | 3 | 0 | No |
| RadQAc [37] | 3 | 1 | Yes |
| EHRSQL [36] | 3 | 0 | Yes |
| Kim et al [39] | 2 | 0 | Yes |
| ClinicalKBQAd [40] | 2 | 0 | No |
| Hamidi and Roberts [48] | 1 | 0 | No |
| MedAlign [49] | 1 | 0 | No |
| RxWhyQA [27] | 0 | 0 | Yes |
| Mishra et al [45] | 0 | 0 | No |
| CLIFTe [47] | 0 | 0 | No |
| Mahbub et al [50] | 0 | 0 | No |
| Dada et al [51] | 0 | 0 | No |
aThis dataset follows the original schema of Medical Information Mart for Intensive Care (MIMIC-III).
bDiSCQ: Discharge Summary Clinical Questions.
cRadQA: Radiology Question Answering Dataset.
dClinicalKBQA: Clinical Knowledge Base Question Answering.
eCLIFT: Clinical Shift.