A Context-based Chatbot Surpasses Trained Radiologists and Generic ChatGPT in Following the ACR Appropriateness Guidelines.

Background Radiological imaging guidelines are crucial for accurate diagnosis and optimal patient care as they result in standardized decisions and thus reduce inappropriate imaging studies. Purpose In the present study, we investigated the potential to support clinical decision-making using an interactive chatbot designed to provide personalized imaging recommendations from American College of Radiology (ACR) appropriateness criteria documents using semantic similarity processing. Methods We utilized 209 ACR appropriateness criteria documents as specialized knowledge base and employed LlamaIndex, a framework that allows to connect large language models with external data, and the ChatGPT 3.5-Turbo to create an appropriateness criteria contexted chatbot (accGPT). Fifty clinical case files were used to compare the accGPT's performance against general radiologists at varying experience levels and to generic ChatGPT 3.5 and 4.0. Results All chatbots reached at least human performance level. For the 50 case files, the accGPT performed best in providing correct recommendations that were "usually appropriate" according to the ACR criteria and also did provide the highest proportion of consistently correct answers in comparison with generic chatbots and radiologists. Further, the chatbots provided substantial time and cost savings, with an average decision time of 5 minutes and a cost of 0.19 € for all cases, compared to 50 minutes and 29.99 € for radiologists (both p < 0.01). Conclusion ChatGPT-based algorithms have the potential to substantially improve the decision-making for clinical imaging studies in accordance with ACR guidelines. Specifically, a context-based algorithm performed superior to its generic counterpart, demonstrating the value of tailoring AI solutions to specific healthcare applications.

A nationwide registry for recurrent urolithiasis in the upper urinary tract - The RECUR study protocol.

BACKGROUND: Urinary stone disease is a widespread disease with tremendous impact on those affected and on societies around the globe. Nevertheless, clinical and health care research in this area seem to lag far behind cardiovascular diseases or cancer. This may be due to the lack of an immediate deadly threat from the disease and therefore less public and professional interest. However, the patients suffer from recurring, sometimes intense pain and often must be treated in hospital. Long-term morbidity includes doubled rates of chronic kidney disease and arterial hypertension after at least one stone-related event. Observational studies, more specifically, registries and other electronic data sets have been proposed as a means of filling critical gaps in evidence. We propose a nationwide digital and fully automated registry as part of the German Ministry for Education and Research (BMBF) call for the "establishment of model registries". METHODS: RECUR builds on the technical infrastructure of Germany's Medical Informatics Initiative. Local data integration centres (DIC) of participating medical universities will collect pseudonymized and harmonized data from respective hospital information systems. In addition to their clinical data, participants will provide patient reported outcomes using a mobile patient app. Scientific data exploration includes queries and analysis of federated data from DICs of eleven participating sites. All primary patient data will remain at the participating sites at all times. With comprehensive data from this longitudinal registry, we will be able to describe the disease burden, to determine and validate risk factors, and to evaluate treatments. Implementation and operation of the RECUR registry will be funded by the BMBF for five years. Subsequently, the registry is to be continued by the German Society of Urology without significant costs for study personnel. DISCUSSION: The proposed registry will substantially improve the structural and procedural framework for patients with recurrent urolithiasis. This includes advanced diagnostic algorithms and treatment pathways. The registry will help us identify those patients who will most benefit from specific interventions to prevent recurrences. The RECUR study protocol and the registry's technical architecture including full digitalization and automation of almost all registry-associated proceedings can be transferred to future registries. TRIAL REGISTRATION: This study is registered at the German Clinical Trial Register (Deutsches Register Klinischer Studien), DRKS-ID DRKS00026923 , date of registration January, 11th 2022.

The H+-ATPase HA1 of Medicago truncatula Is Essential for Phosphate Transport and Plant Growth during Arbuscular Mycorrhizal Symbiosis.

A key feature of arbuscular mycorrhizal symbiosis is improved phosphorus nutrition of the host plant via the mycorrhizal pathway, i.e., the fungal uptake of Pi from the soil and its release from arbuscules within root cells. Efficient transport of Pi from the fungus to plant cells is thought to require a proton gradient across the periarbuscular membrane (PAM) that separates fungal arbuscules from the host cell cytoplasm. Previous studies showed that the H+-ATPase gene HA1 is expressed specifically in arbuscule-containing root cells of Medicago truncatula. We isolated a ha1-2 mutant of M. truncatula and found it to be impaired in the development of arbuscules but not in root colonization by Rhizophagus irregularis hyphae. Artificial microRNA silencing of HA1 recapitulated this phenotype, resulting in small and truncated arbuscules. Unlike the wild type, the ha1-2 mutant failed to show a positive growth response to mycorrhizal colonization under Pi-limiting conditions. Uptake experiments confirmed that ha1-2 mutants are unable to take up phosphate via the mycorrhizal pathway. Increased pH in the apoplast of abnormal arbuscule-containing cells of the ha1-2 mutant compared with the wild type suggests that HA1 is crucial for building a proton gradient across the PAM and therefore is indispensible for the transfer of Pi from the fungus to the plant.

Cell-specific protein profiling in Arabidopsis thaliana trichomes: identification of trichome-located proteins involved in sulfur metabolism and detoxification.

Metabolite, protein, and transcript analysis at the cellular level gives unparalleled insight into the complex roles tissues play in the plant system. However, while capillary electrophoresis and PCR amplification strategies make the profiling of metabolites and transcripts in specific cell types possible, the profiling of proteins in small samples represents a bottleneck. Here for the first time protein profiling has been achieved in a specific plant cell type: The application of specific cell sampling and shotgun peptide sequencing (nano LC/MS/MS) resulted in the identification of 63 unique proteins from pooled Arabidopsis trichome cells. A complete S-adenosylmethionine pathway cluster, two S-adenosylmethionine synthase isoforms, a glutathione S-conjugate translocator and other proteins involved in sulfur metabolism and detoxification are shown to be present in these cells, in agreement with previous work done at the level of trichome transcript analysis. The technology described here brings the simultaneous identification and localization of physiologically relevant cellular proteins within reach.

Quantitative expression analysis of selected transcription factors in pavement, basal and trichome cells of mature leaves from Arabidopsis thaliana.

Gene expression levels of several transcription factors from Arabidopsis thaliana that were described previously to be involved in leaf development and trichome formation were analysed in trichome, basal and pavement cells of mature leaves. Single cell samples of these three cells types were collected by glass micro-capillaries. Real-time reverse transcription (RT)-PCR was used to analyse expression patterns of the following transcription factors: MYB23, MYB55, AtHB1, FILAMENTOUS FLOWER (FIL)/YABBY1 (YAB1), TRIPTYCHON (TRY) and CAPRICE (CPC). A difference in the expression patterns of TRY and CPC was revealed. Contrary to the CPC expression pattern, no transcripts of TRY could be detected in pavement cells. FIL/YAB1 was exclusively expressed in trichome cells. AtHB1 was highly expressed throughout all three cell types. MYB55 was higher expressed in basal cells than in trichome and pavement cells. MYB23 showed a pattern of low expression in pavement cells, medium in basal cells and high expression in trichomes. Expression patterns obtained by single cell sampling and real-time RT-PCR were compared to promoter GUS fusions of the selected transcription factors. Therefore, we regenerated two transgenic Arabidopsis lines that expressed the GUS reporter gene under control of the promoters of MYB55 and YAB1. In conclusion, despite their function in leaf morphogenesis, all six transcription factors were detected in mature leaves. Furthermore, single cell sampling and promoter GUS staining patterns demonstrated the predominant presence of MYB55 in basal cells as compared to pavement cells and trichomes.

Gene expression profiling of single epidermal, basal and trichome cells of Arabidopsis thaliana.

Samples of single epidermal, basal and trichome cells were collected by glass microcapillaries from 7-week-old Arabidopsis thaliana leaves. Transcript amplification of these single-cell samples was performed by RT PCR. For gene expression profiling, we hybridized the amplified transcriptome of each individual cell type to nylon membranes spotted with 16,000 Arabidopsis expressed sequence tags (ESTs). Initial analysis of the array filter data enabled us to functionally categorize transcripts that were present in each individual cell type. In order to confirm the filter array data, we used RT PCR. Results of this RT PCR approach confirmed the presence of 12 selected candidate genes in agreement with array filter hybridization data. Further, transcripts involved in detoxification and sulfur metabolism could be identified in epidermal cell extracts. Together, the results of our study provide the localization of approximately 1000 expressed genes to either pavement, basal or trichome cells. To cluster transcripts with similar expression levels, we developed a novel mathematical algorithm. Based on the mean and standard deviation, ratios of expression levels of a transcript were defined for pairs of the three cell types. This numerical analysis enabled subdivision into 67 categories of genes differentially expressed in epidermal, basal and trichome cells. Transcripts in each category displayed similar ratios of expression levels in the three cell types. Examples of these clusters are presented and discussed in Appendix A.