Cody, the AI That’s Revolutionizing Medical Coding
How can a machine understand a patient record and code it accurately? Swisscoding Technologies has taken up this challenge with the support of Cody, an AI that works like a junior coder. But how does it work?
After a year of development, based on its own expertise in medical coding and hundreds of thousands of coded documents, Swisscoding Technologies is launching Cody, a new, fully integrated technology product capable of independently handling the medical coding of patient records.
A response to the requirements of the field
Cody is a tool that evolves rapidly thanks to its technical SaaS design («software as a service»), which is improved almost daily. Swisscoding has the largest medical coding team in Switzerland, reinforced by AI developers and engineers, making it ideally positioned to develop a fully automated coding service based on AI. Through six years of collaboration with numerous clinics and hospitals, the company understands the specific needs of institutions and coders and possesses – thanks to the collaboration of selected clients – large amounts of data to effectively train the AI. This unique combination of expertise ensures that the designed solution fully meets the industry’s requirements.
Scalable Learning
Cody learns like a young medical coder. It learns by coding, being corrected, and comparing its results with the «correct» results. Swisscoding has access to hundreds of thousands of cases that have been correctly coded, reviewed by external auditors, incorrectly coded, corrected, and supplemented. This learning ability allows Cody to adapt and improve by identifying recurring errors and optimization opportunities. A particularly innovative approach by Swisscoding involves pitting two AI systems against each other: one that reproduces previous results and another that identifies areas for improvement. This strategy leads to more reliable and accurate coding results than those achieved exclusively through human methods or other tools available on the market, while also preparing for significant developments in the coming years.
«If the coding is deemed unreliable by the AI, the file is managed manually. Only error-free codings are directly integrated into the hospital’s system. The goal is to reduce the workload of coders, not to burden them with new tasks such as validations or additional checks»
Christophe Rosso, CIO of Swisscoding Technologies
Self-Evaluation as a Guarantee of Reliability
A key aspect of AI learning is its ability to self-evaluate. From the very beginning of the tool development, Swisscoding decided to go beyond simple code suggestions, as these solutions extend coding time and require human intervention. The idea is to produce a final result that can be used directly for invoicing without requiring revision. The tool must be able to decide for each individual case whether it can code independently with the expected quality or whether it is better for a human to intervene. Using the technology employed by Swisscoding, confidence intervals – or «confidence levels» – were incorporated. These levels allow Cody to measure in real-time how confident it is in each case. This system enables it to self-assess and adjust its interventions to the complexity and requirements of each situation.
Robot Portrait
Cody is an innovative solution for medical coding, powered by AI and developed by Swisscoding Technologies. The result of a unique synergy between medical coding experts and AI engineers, it stands out with its numerous features:
- It codes simple cases with the same or even higher reliability than human coders.
- It is multilingual and geographically unlimited, allowing it to adapt to all environments.
- It seamlessly integrates the advancements of SwissDRG, ensuring continuous compliance.
- Without visible interfaces, it integrates into existing systems, making itself unobtrusive and autonomous.
- Thanks to its ability to learn and self-evaluate, it delivers only flawless results.
- It can be used during the hospital stay before coding to support medical staff and case managers.
- It adheres to the highest security standards to ensure data protection.
- It supports a transformative vision of the hospital administration system.
- Finally, it also simplifies the work of insurance companies by reducing their control burden.
No Interface, No Additional Work
Cody has no user interface. Unlike other solutions that require the addition of new software and implementation processes, this tool integrates directly into the existing systems of hospitals. It retrieves the data where it is stored and automatically feeds it into the invoicing tools without requiring human management or control. This eases coders without burdening them with additional tasks. If Cody cannot guarantee flawless quality in a case, it does not intervene, thereby avoiding double work. This invisible but efficient way of working ensures a smooth and integrated automatic workflow.
Towards Iterative Coding During Hospital Stay
Cody not only revolutionizes medical coding for invoicing but can also be used during the hospital stay. With Cody, the information collected in the clinical information system (CIS) can be used throughout the stay to anticipate and correct inaccuracies in medical documentation. For example, thanks to the automatically updated coding after each new entry into the CIS, Cody predicts the most relevant DRG based on the existing data and can thus identify if information is incomplete for optimal coding. In this case, the tool asks targeted questions to document it immediately. This helps not only the coders but also the medical and nursing staff with administrative care.
At the same time, AI offers powerful forecasting tools, such as the ability to predict the patient’s discharge time. Upon admission, it analyzes the first available data, such as the type of procedure or the initial diagnosis, to make a preliminary estimate of the length of stay. This prediction is refined as new information flows in throughout the stay. This not only facilitates the planning of hospital resources but also helps the case manager anticipate logistical and administrative requirements.
«It’s not just about trusting AI, but about engaging in a dialog to understand fears, adapt solutions and collaboratively create tools that truly meet the needs of hospitals.» Stefan Stefaniak, CEO of Swisscoding
Benefits Across the Board
The integration of AI in medical coding could also mean relief for insurers. Through more transparent and objective invoicing and optimized controls in advance, insurers would need to conduct fewer checks, significantly reducing administrative costs that increase premiums without adding value to patient care. Swisscoding Technologies is working on developing certified and auditable processes that would ultimately not replace legal processes but would allow certain invoices to be validated directly by insurers without further checks. Although this model still needs time to be fully implemented, it offers a promising vision for simplifying interactions between hospitals and insurers.
A near future
The vision remains the same: to transform the organization of the hospital by drastically reducing administrative burdens, allowing doctors and nursing staff to fully focus on their patients. We envision a hospital where most of the time is spent on treatment rather than filling out paperwork. With the help of AI, Swisscoding has found an effective ally that promises to put patients back at the center of hospital priorities.
Related Posts
Meet the team Cintia
We’re delighted to introduce Cintia Maret, who joined Swisscoding in January 2025 as a Documentation Operator !
Meet the team Anna
We are delighted to welcome Anna Waxweiler, who joins Swisscoding in March 2025 to strengthen our Medizincontrolling team in German-speaking Switzerland.
