A Lung Cancer Detection Desktop Application for a Biotechnology Company

Summary

INNERLUXES partnered with a US-based biotechnology company and, in two months, delivered a desktop app for lung cancer detection. The app ensures 100% stable generation of comprehensive diagnostic reports based on the analysis of flow cytometry data.

About the Client

The Client is a US-based biotechnology company that develops diagnostic tests to identify early-stage lung cancer and other lung diseases, such as COPD and asthma, using flow cytometry technology.

The Client wanted to accelerate the market entry of its software product and needed an ISO 13485-certified development partner for its lung cancer detection medical device software. The Client had developed a proprietary R-based ML algorithm used in the validation trial of a noninvasive test for the early detection of lung cancer, and needed help transforming the algorithm into a full-fledged desktop analytics application.

Design of a Lung Cancer Detection Solution

The Client entrusted INNERLUXES with the project because of its ISO-certified quality management system and expertise in FDA 510(k) and CE marking submissions. Within a week, INNERLUXES assembled a team of a full-stack developer/solution architect and a project manager. Since the Client had a crystal-clear vision of the desired project scope and rigid software requirements, INNERLUXES's PM suggested following the waterfall development methodology.

The discovery stage comprised two steps:

• Requirements engineering. The solution architect studied the Client's requirements, desired functionality, and the existing flow cytometry R script, and discussed the specifics of the algorithm with the Client's technical and scientific team.
• Software architecture and tech stack planning. With sufficient project information at hand, the solution architect planned the architecture and technical stack of the desktop app. He chose to write the core business logic in C++, as this language is optimal for intense data processing and complex calculations. The app would use the R language to communicate with the analytics script and HL7 to securely process patient information from a laboratory information system (LIS).

Development of the Lung Cancer Detection Software Within Two Months

In just two months after the healthcare software concept was approved, INNERLUXES delivered the desktop application. The diagnostic system using the new app functions as follows:

• The lung cancer detection app runs on a laboratory computer connected to the flow cytometer. The cytometer analyzes cellular events in a patient's sputum sample and sends a set of output files to a secure network drive available from the laboratory computer.
• The desktop app inserts the path to the output files into the cancer detection R script.
• The desktop app receives an HL7 message with the necessary patient data (for example, age) from a laboratory information system database. This data informs the diagnostic results, so it is essential to consider it during the analysis.
• The HL7 message with patient data is transformed into R script input arguments.
• The desktop app runs the data analysis script and interprets the lung cancer diagnostic results.
• The generated results are then sent as an HL7 message to the LIS.

INNERLUXES's regulatory compliance consultant created technical documentation for the lung cancer detection software following the requirements for IVDR (In Vitro Diagnostics Regulation) submission.

After user acceptance testing, the Client was satisfied with the desktop app, and INNERLUXES proceeded with its deployment.

Key Value for the Client

• Desktop software developed in just two months.
• Technical documentation ready for IVDR submission.
• 100% stable generation of comprehensive diagnostic reports.
• The Client's diagnostic solution showed 92% sensitivity and 87% specificity in detecting lung cancer in high-risk patients with small pulmonary nodules under 20 millimeters, and 82% sensitivity and 88% specificity across all nodule sizes.

Technologies and Regulatory Requirements

C++, R, HL7, Docker, IVDR.