Health Department
Decision Support System for Optimizing Hospital Admissions
Industry: Healthcare & Life Sciences
The Health Department operates several acute care clinics and specialized units focusing on internal medicine, oncology and geriatrics
As part of a strategic digitization initiative, the goal was to improve the allocation of medical resources, enhance admission decisions and ensure high-quality patient care. Capacity limitations in inpatient care had to be addressed through data-driven support – without burdening staff or interfering with existing clinical documentation systems.
Status quo
Prior to the project, many inpatient admissions were based on individual judgment without digital, evidence-based support.
There was no consistent evaluation of whether an admission was medically necessary, appropriate or avoidable. Especially in complex cases or when free-text documentation was involved, reliable criteria for decision-making were lacking. This often led to avoidable admissions, inefficient bed usage and increased workload for both clinical and administrative teams.
The challenge
The main challenge was to build a lean and clinically viable system capable of analyzing free-text data from the admission process in real time while fully complying with data protection regulations.
The tool had to identify relevant indicators, previous stays and care patterns based solely on text data – without requiring additional interfaces or disrupting clinical workflows. The ultimate goal was to deliver automated, transparent recommendations for hospital admissions.
The solution
CONVOTIS developed a modular decision support system based on natural language processing, decision trees and large language models with medical vocabulary.
Patient records from the past 20 months were analyzed in a strictly separated training environment. The system processes clinical free-text inputs, detects patterns such as readmissions or prior emergency visits and delivers automated recommendations on hospital admission or alternative care – seamlessly integrated into the clinical context, with no additional logins required.
The result
The solution demonstrably reduces unnecessary hospital admissions – particularly in cases lacking a clear medical indication.
Bed occupancy was measurably optimized, resources could be planned more effectively, and patient pathways were streamlined. Physicians receive a solid basis for admission decisions without being disrupted in their routine. At the same time, hospital management benefits from more accurate case figures and improved budget control – resulting in measurable gains for staff, operations and patient care.
Technical
Highlights
Even complex projects can be summarised in clear technical terms. Key figures and distinctive features provide insight into the concrete implementation – measurable, tangible, and transparent.
- Analysis of 20 months of patient data
- Classification via decision trees and neural networks
- Integration of LLMs with medical tokenization
- Real-time processing of free-text admission data
- Clear separation of training and production environments
- Indication-based recommendation logic
The system supports medical staff during hospital admissions without disrupting workflows. Resources are used more effectively, bed occupancy is reduced and unnecessary treatments are avoided – delivering measurable value for staff, patients and hospital operations.
Your IT Transformation starts here.
Let’s talk about your goals.
You’ve seen our success stories – now let’s plan your IT transformation together. Whether you are at the beginning of your digital journey or ready for the next step, our experts are here to support you.