Tailored to Fit Sepsis Individuals: Medical Knowledge Aware Reinforcement Learning Model Offers Optimized Therapeutic Strategies
Bibliographic record
Abstract
Sepsis is a life-threatening syndrome, with high morbidity and mortality. Timely treatments and precise interventions are crucial for improving sepsis patient outcomes. Reinforcement learning (RL) models have made promising advances in associating sepsis treatments. However, existing models face obstacles when applied to heterogeneous sepsis patients with an imbalanced distribution of disease severity, struggling to maintain optimal performance. To provide optimal therapeutic intervention strategies for sepsis with different disease severity, we proposed a novel treatment recommendation model named Medical ontology knowledge and disease diagnosis Position aware Transfer learning - Double Dueling Deep Q Network (MPT-D3QN). Through the attention mechanism, medical ontology knowledge and disease diagnosis positions were introduced to achieve a more accurate representation of patient states. Subsequently, an offline deep reinforcement learning algorithm was used to recommend therapeutic intervention strategies for a specific patient population. The transfer learning framework was responsible for the information transfer across patient groups in different domains. We trained and tested MPT-D3QN model on the Multi-Parameter Intelligent Monitoring in Intensive Care III (MIMIC-III) dataset. Compared with the reported strategies in present clinical practice, the MPT-D3QN model could obtain a higher expected return value (12.98 vs. 10.61) and reduce the estimated mortality from 13.20% to 6.15% in all test datasets. Moreover, the experimental test on an external dataset eICU Collaborative Research Database (eICU) further demonstrated the robust generalization capability of the model. Compared with the state-of-the-art models, the proposed model not only guaranteed higher expected returns but also generated optimal and interpretable treatment strategies for sepsis with different disease severity.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.002 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.000 |
| Bibliometrics | 0.001 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.001 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one teacher head, not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".