Real-time optimization of diagnostic processes through digitalization

^{Mohamed Hassan / Pixabay}

The DiagnoStream project addresses a data challenge of the IOBM (Institute of Osteology and Biomechanics) at the University Medical Center Hamburg-Eppendorf, which treats over 12,000 patients a year. The aim of the project is to develop tailor-made management software that manages and optimizes the daily patient flows. The project includes the use of hardware to record process data, the algorithmic optimization of operating procedures and the development of a graphical user interface for medical staff.

The challenge: 

To date, there is no suitable IT system for planning operational processes in the IOBM. The tasks that arise are managed analogously using the physical patient file. To do this, the doctors tick off on a checklist which examinations are due for a patient. All patient files are stored on a shelf with their checklists and are processed by the medical assistants (MFA) according to the first-in, first-out (FIFO) principle. An MFA takes out the patient file and carries out one or more of the pending examinations. Since the process is analog, it can happen that the MFA unknowingly takes the patient to a room that is already occupied. In some cases, this leads to the examination being aborted and the MFA sends the patient back to the waiting room. Her file is then returned to the back of the queue according to the FIFO principle. The patient now has to wait again until all the files in front of her have been processed once. This regularly leads to long throughput and waiting times. Another problem on busy days is that there is no system that provides information about the current waiting times of patients. This makes it difficult to prioritize patients fairly.

From a management perspective, there is also a lack of reliable statistics on waiting times, throughput times, examination times, equipment utilization, staff utilization and much more. A system is therefore needed that prioritizes the pending examinations efficiently and fairly and provides management with sufficient process data for analysis.

Solution approach:

The required system should record all relevant process data live and algorithmically calculate the optimum operating sequence. Information on available staff, patient waiting times, upcoming examinations, room occupancy and routes between examination rooms should be included in the optimization. Together with other useful data, this optimized queue will be made available to staff in an interactive GUI. This data-driven solution enables staff to work with maximum efficiency without having to worry about the order of patients. As the system learns over time, the predictions for turnaround times become more accurate. Additional features will enable the export of data, the creation of analyses and manual adjustments to the queue order.

Data concept:

Each patient's upcoming examinations are recorded manually via a web application. The remaining process data is recorded using barcode scanners installed in each room. A barcode on the patient file is read by the scanner both when entering and leaving the room. In this way, the system knows when which examination takes place in which room. By combining and calculating this data, the entire patient journey can be calculated. This also forms the basis for prioritizing patients and calculating key figures for management.

Hardware and technologies used:

• Hardware: Raspberry Pis and barcode scanner
• Database: MySQL
• Programming language: Python

And here the group in their own words (German):