| {
|
| "meta_info": {
|
| "case_id": "MentalArithmetic_01.edf",
|
| "bench_subset": "NeuroBench-Core",
|
| "difficult": 1.5,
|
| "original_dataset": "MentalArithmetic"
|
| },
|
| "agent_input": {
|
| "data_path": "data/core/MentalArithmetic_01.edf",
|
| "instruction": "Please determine which lead-distribution system this recording follows (10-20, 10-10 or 10-5), and then list all channels that do not belong to standard measurement electrodes of that system. Return the non-standard channels using their original channel names exactly as they appear in the data. If all channels belong to standard measurement electrodes, return an empty list."
|
| },
|
| "eval_config": {
|
| "parser_prompt": "### ROLE\nYou are a precise JSON Extraction Engine for neurophysiological data. Your sole task is to convert natural language reports into structured JSON data.\n\n### TASK\nExtract two outputs from the agent report:\n1) lead system classification\n2) final list of channels that are not standard measurement electrodes of that system\n\n### STRICT CONSTRAINTS (MANDATORY)\n1. Output ONLY a valid JSON object.\n2. DO NOT include Markdown code blocks.\n3. DO NOT include conversational text or explanations.\n4. The keys in JSON must be EXACTLY \"lead_system\" and \"non_standard_channels\".\n5. \"lead_system\" must be exactly one of \"10-20\", \"10-10\", \"10-5\", or null.\n6. Extract the explicitly reported final non-standard-channel list, not the full channel inventory or the standard-channel list.\n7. Normalize each returned display name only to the benchmark naming contract: trim whitespace and remove one leading acquisition prefix EEG followed by a space, hyphen, or underscore.\n8. If a modality prefix is immediately duplicated, collapse it to one label; for example, \"ECG ECG\" becomes \"ECG\". Preserve the remaining spelling, case, punctuation, and reference identity exactly.\n9. Do NOT remove terminal reference parts and do NOT convert legacy aliases for this case. Required examples: \"EEG A2-A1\" -> \"A2-A1\"; \"EEG 23A-23R\" -> \"23A-23R\"; \"ECG ECG\" -> \"ECG\"; \"SaO2\" -> \"SaO2\".\n10. If the report explicitly says no non-standard channel exists, return an empty array []. If the final list is missing or unclear, set non_standard_channels to null.\n11. If lead_system is missing or unclear, set only lead_system to null; do not guess either value.\n\n### OUTPUT TEMPLATE\n{\"lead_system\": <\"10-20\"|\"10-10\"|\"10-5\"|null>, \"non_standard_channels\": <array|null>}",
|
| "metrics": [
|
| {
|
| "metric_id": "lead_system_accuracy",
|
| "type": "categorical_check",
|
| "target_key": "lead_system",
|
| "weight": 50,
|
| "params": {
|
| "gt_value": "10-20"
|
| }
|
| },
|
| {
|
| "metric_id": "non_standard_channels_match",
|
| "type": "set_match_check",
|
| "target_key": "non_standard_channels",
|
| "weight": 50,
|
| "params": {
|
| "gt_value": [
|
| "A2-A1",
|
| "ECG"
|
| ],
|
| "match_mode": "exact"
|
| }
|
| }
|
| ]
|
| }
|
| } |
