Medical Plan Generation on In-house Clinical EHR Dataset
0.3183BLEUMEDPLAN (Medical-Mixtral-7B-v2k)
Evaluation results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| MEDPLAN (Medical-Mixtral-7B-v2k)Planning Method=S+O->A->P, Self-history=true, Instruction Tuning=true, Cross-patient=true, Backbone=Medical-Mixtral-7B-v2k2025.03 | 0.3183 | 0.5213 | 0.5817 | 0.4758 | 0.5771 | 0.9172 | |
| MEDPLAN (Medical-Llama3-8B)Planning Method=S+O->A->P, Self-history=true, Instruction Tuning=true, Cross-patient=true, Backbone=Medical-Llama3-8B2025.03 | 0.3147 | 0.5162 | 0.5761 | 0.4696 | 0.5712 | 0.9155 | |
| Medical-Mixtral-7B-v2kPlanning Method=S+O->P, Self-history=true, Instruction Tuning=true, Cross-patient=true2025.03 | 0.3124 | 0.5108 | 0.5703 | 0.4649 | 0.5658 | 0.9142 | |
| MEDPLAN (Bio-Medical-Llama3-8B)Planning Method=S+O->A->P, Self-history=true, Instruction Tuning=true, Cross-patient=true, Backbone=Bio-Medical-Llama3-8B2025.03 | 0.3122 | 0.5167 | 0.5748 | 0.4675 | 0.5697 | 0.915 | |
| Bio-Medical-Llama3-8BPlanning Method=S+O->P, Self-history=true, Instruction Tuning=true, Cross-patient=true2025.03 | 0.3095 | 0.5015 | 0.5579 | 0.4568 | 0.5539 | 0.9116 | |
| Medical-Llama3-8BPlanning Method=S+O->P, Self-history=true, Instruction Tuning=true, Cross-patient=true2025.03 | 0.3074 | 0.5014 | 0.5592 | 0.4566 | 0.5544 | 0.9117 | |
| Bio-Medical-Llama3-8BPlanning Method=S+O->P, Self-history=true, Instruction Tuning=true, Cross-patient=false2025.03 | 0.2994 | 0.4866 | 0.5442 | 0.4417 | 0.5396 | 0.9088 | |
| Medical-Mixtral-7B-v2kPlanning Method=S+O->P, Self-history=true, Instruction Tuning=true, Cross-patient=false2025.03 | 0.2983 | 0.483 | 0.5418 | 0.4427 | 0.5378 | 0.9105 | |
| Medical-Llama3-8BPlanning Method=S+O->P, Self-history=true, Instruction Tuning=true, Cross-patient=false2025.03 | 0.2912 | 0.4773 | 0.5353 | 0.4342 | 0.5311 | 0.9078 | |
| Medical-Llama3-8BPlanning Method=S+O->P, Self-history=false, Instruction Tuning=true, Cross-patient=false2025.03 | 0.1786 | 0.3066 | 0.3434 | 0.2749 | 0.3402 | 0.8673 | |
| Medical-Mixtral-7B-v2kPlanning Method=S+O->P, Self-history=false, Instruction Tuning=true, Cross-patient=false2025.03 | 0.1703 | 0.3116 | 0.3653 | 0.2852 | 0.3605 | 0.87 | |
| Bio-Medical-Llama3-8BPlanning Method=S+O->P, Self-history=false, Instruction Tuning=true, Cross-patient=false2025.03 | 0.1128 | 0.217 | 0.2358 | 0.1741 | 0.2309 | 0.8484 | |
| Medical-Mixtral-7B-v2kPlanning Method=S+O->P, Self-history=true, Instruction Tuning=false, Cross-patient=false2025.03 | 0.0672 | 0.1966 | 0.2497 | 0.1561 | 0.2435 | 0.8522 | |
| Bio-Medical-Llama3-8BPlanning Method=S+O->P, Self-history=true, Instruction Tuning=false, Cross-patient=false2025.03 | 0.0613 | 0.1881 | 0.2351 | 0.1481 | 0.2287 | 0.85 | |
| Medical-Llama3-8BPlanning Method=S+O->P, Self-history=true, Instruction Tuning=false, Cross-patient=false2025.03 | 0.0528 | 0.1734 | 0.22 | 0.1296 | 0.2145 | 0.8475 | |
| GPT-4oPlanning Method=S+O->P, Self-history=true, Instruction Tuning=false, Cross-patient=false2025.03 | 0.0288 | 0.1663 | 0.1541 | 0.0702 | 0.1396 | 0.827 | |
| o1Planning Method=S+O->P, Self-history=true, Instruction Tuning=false, Cross-patient=false2025.03 | 0.0164 | 0.1404 | 0.1254 | 0.0464 | 0.1079 | 0.8171 |