Few-Shot and Prompt Training for Text Classification in German Doctor's Letters.

To classify sentences in cardiovascular German doctor's letters into eleven section categories, we used pattern-exploiting training, a prompt-based method for text classification in few-shot learning scenarios (20, 50 and 100 instances per class) using language models with various pre-training approaches evaluated on CARDIO:DE, a freely available German clinical routine corpus. Prompting improves results by 5-28% accuracy compared to traditional methods, reducing manual annotation efforts and computational costs in a clinical setting.

A distributable German clinical corpus containing cardiovascular clinical routine doctor's letters.

We present CARDIO:DE, the first freely available and distributable large German clinical corpus from the cardiovascular domain. CARDIO:DE encompasses 500 clinical routine German doctor's letters from Heidelberg University Hospital, which were manually annotated. Our prospective study design complies well with current data protection regulations and allows us to keep the original structure of clinical documents consistent. In order to ease access to our corpus, we manually de-identified all letters. To enable various information extraction tasks the temporal information in the documents was preserved. We added two high-quality manual annotation layers to CARDIO:DE, (1) medication information and (2) CDA-compliant section classes. To the best of our knowledge, CARDIO:DE is the first freely available and distributable German clinical corpus in the cardiovascular domain. In summary, our corpus offers unique opportunities for collaborative and reproducible research on natural language processing models for German clinical texts.

Automatic extraction of 12 cardiovascular concepts from German discharge letters using pre-trained language models.

OBJECTIVE: A vast amount of medical data is still stored in unstructured text documents. We present an automated method of information extraction from German unstructured clinical routine data from the cardiology domain enabling their usage in state-of-the-art data-driven deep learning projects. METHODS: We evaluated pre-trained language models to extract a set of 12 cardiovascular concepts in German discharge letters. We compared three bidirectional encoder representations from transformers pre-trained on different corpora and fine-tuned them on the task of cardiovascular concept extraction using 204 discharge letters manually annotated by cardiologists at the University Hospital Heidelberg. We compared our results with traditional machine learning methods based on a long short-term memory network and a conditional random field. RESULTS: Our best performing model, based on publicly available German pre-trained bidirectional encoder representations from the transformer model, achieved a token-wise micro-average F1-score of 86% and outperformed the baseline by at least 6%. Moreover, this approach achieved the best trade-off between precision (positive predictive value) and recall (sensitivity). CONCLUSION: Our results show the applicability of state-of-the-art deep learning methods using pre-trained language models for the task of cardiovascular concept extraction using limited training data. This minimizes annotation efforts, which are currently the bottleneck of any application of data-driven deep learning projects in the clinical domain for German and many other European languages.

Temporary oral anticoagulation after MitraClip - a strategy to lower the incidence of post-procedural stroke?

Background: Incidence of stroke within 30 days after percutaneous mitral valve repair using the MitraClip varies from 0.7% and 2.6% between registries. Standard medical treatment after the MitraClip procedure, in the absence of risk factors requiring antithrombotic therapy such as atrial fibrillation, is dual antiplatelet therapy using aspirin and clopidogrel. ESC/EACTS and ACC/AHA surgical guidelines show a Class IIa indication for temporary antithrombotic therapy after mitral valve repair/bioprosthetic valve replacement within the first three months even in patients with no additional risk factors.Methods: 470 patients were treated with the MitraClip receiving oral anticoagulation (Coumadin: INR 2.0-3.0) instead of dual antiplatelet therapy for at least 30 days after the procedure. The incidence of stroke, as well as major adverse events such as bleeding, were analysed in comparison to large registries and multi-centre studies.Results: Incidence of stroke within 30 days was significantly reduced as compared to comparative cohorts (0.2% vs. Median 1.3% [0.7-2.6%]; p < .05). Cardiovascular risk factors such as age, atrial fibrillation, hypertension, diabetes, STS score and prior stroke did not differ significantly between cohorts (ns). Bleeding complications were not increased due to 30 days oral anticoagulation treatment (4.6% vs. Median 7.4% [3.4-13.6%]; ns).Conclusions: Temporary oral anticoagulation might be a strategy to reduce the incidence of stroke within the first 30 days after the MitraClip procedure. Bleeding events were not significantly altered due to temporary oral anticoagulation.

NOAC monotherapy in patients with concomitant indications for oral anticoagulation undergoing transcatheter aortic valve implantation.

AIMS: Antithrombotic therapy after transcatheter aortic valve implantation (TAVI) is highly controversial and guideline recommendations are not evidence based. We assessed efficacy and safety of non-vitamin K antagonist oral anticoagulant (NOAC) monotherapy in patients with concomitant indications for OAC undergoing TAVI. METHODS: Among more than 1300 TAVI patients since 2008, 154 patients were identified who received postinterventional NOAC monotherapy. Outcomes were compared to 172 patients treated with vitamin K antagonist (VKA) monotherapy. Thromboembolic as well as bleeding complications were analysed for 6 months after TAVI. RESULTS: Despite high CHA2DS2-Vasc (4.6 ± 1.2), HEMORR2HAGES (4.7 ± 1.9) and HASBLED (2.7 ± 0.8) scores only three major/life-threatening bleedings and four thromboembolic events occurred after NOAC therapy had been initiated post-TAVI. 12 patients (7.8%) died within 6 months after the procedure. Compared to patients being treated with VKA monotherapy, analysis of a combined end-point of post-procedural death, stroke, embolism and severe bleeding revealed no significant differences (17/154 vs. 14/172; p = 0.45). CONCLUSIONS: The results of this study suggest that NOAC therapy without additional antiplatelet treatment is effective and safe in patients with concomitant indications for OAC undergoing TAVI.

Feasibility and safety of vitamin K antagonist monotherapy in atrial fibrillation patients undergoing transcatheter aortic valve implantation.

AIMS: We aimed to assess the efficacy and safety of vitamin K antagonist (VKA) monotherapy in atrial fibrillation (AF) patients undergoing transcatheter aortic valve implantation (TAVI). METHODS AND RESULTS: In 735 TAVIs since 2008 we identified 167 patients suffering from concomitant AF who received either VKA monotherapy (n=77), VKA plus single antiplatelet therapy (SAPT, n=41) or a triple anticoagulation regimen (n=49). Thromboembolic as well as bleeding complications were analysed for six months after TAVI. Only one minor bleeding and no thromboembolic events occurred after VKA therapy had been initiated post TAVI. Compared to patients being treated with additional either single or dual antiplatelet therapy, the incidence of major/life-threatening bleeding complications was significantly lower in the VKA mono group (0/77 [VKA mono] vs. 3/41 [VKA+SAPT; p=0.04] vs. 4/49 [triple anticoagulation; p=0.02]). Analysis of a combined endpoint of post-procedural death, stroke, embolism and major bleeding revealed a significant superiority of VKA monotherapy compared to VKA plus SAPT or DAPT, respectively (5/77 vs. 9/41 [p=0.02] vs. 14/49 [p=0.002]). CONCLUSIONS: VKA therapy without additional antiplatelet treatment is effective and safe in AF patients undergoing TAVI.