How GenAI can improve

quality assurance processes

for Healthcare solutions?
Introduction
In healthcare, medical software and applications are fundamental
for managing and facilitating daily interactions and administrative
duties for an ecosystem of patients, medical professionals, and
health executives. With digital health being the new norm, the
market for medical software is only expected to grow [1] – making
Quality Assurance (QA) integral for ensuring the reliability,
functionality, and quality of these platforms.

There’s scope for innovating how enterprises approach QA across

the Software Testing Life Cycle (STLC) by leveraging emerging
technologies like Generative AI (GenAI). The ultimate objective is to
save time, money, and effort while achieving accuracy, improving
patient-centricity, speeding time to market, and ensuring
compliance adherence.

IDC’s bold prediction for the APEJ region indicates that by 2028,
GenAI tools will be expected to write a whopping 70% of software
tests. This can potentially revolutionize the testing landscape, not
just by automating tasks but also by significantly improving test
coverage, software usability, and even code quality.[2]

1
Generative AI
possibilities for QA
Software testing accounts for 50-60% of the total cost of software
development including data verification, functional testing, automated
verification, and others.[3] GenAI can help improve the QA process and
reduce the overall cost by managing resources efficiently, finding
patterns, spotting unusual activity, and predicting potential problems.

Compared to the traditional testing process, GenAI's cognitive approach

facilitates predictive analysis, empowering developers and QA teams to
enact preventative measures and enhance the software's overall quality.

Following are some areas where GenAI helps drive better outcomes.

Synthetic data generation (SDG)

GenAI equips QA teams to create diverse, representative datasets resembling real-world scenarios for unbiased testing and zero ethical
dilemmas associated with real patient data. Using synthetic data for software quality assurance has multiple applications, such as
generating diverse test cases to assess the software's behavior, performance, and resilience to deviations. This proactive approach helps
detect deviations before they become more significant problems in the early testing phases.

It also fosters a culture of continuous innovation by enabling continuous testing, especially when real-time data is required. Incidentally,
Gartner predicts that 60% of AI data[4] in 2024 will be synthetic, aimed at simulating reality or projecting future scenarios.

2
Scenario exploration
QA teams can explore a broad spectrum of potential scenarios with GenAI. They can uncover edge cases and rare clinical situations that
might have been overlooked during manual testing with fixed datasets. GenAI can also create complex scenarios that mimic real-world
usage patterns, helping QA teams evaluate how the software responds to different user inputs and system states.

Additionally, GenAI can intelligently prioritize scenarios based on factors such as criticality, frequency of occurrence, or impact on the
system. This helps QA teams focus on exploring high-priority scenarios first, ensuring thorough testing of the software's most important
features.

Continuous monitoring and analysis

While traditional AI relies on defined rules and human-programmed instructions, GenAI is more dynamic. In addition to creating test cases
based on evolving requirements, it allows continuous monitoring and analysis to detect anomalies and potential bugs in the software
automatically.

GenAI establishes a continuous feedback loop, learning from the results of previous tests and interactions to adjust its testing strategies
accordingly. This self-learning capability improves its adaptability over time as it becomes more efficient in detecting vulnerabilities and
areas for improvement in the software.

Complex system testing

GenAI excels at simulating real-world scenarios and exploring edge cases for complex testing. The software interacts with various
components, systems, and environments during simulation-based tests, including complex user workflows, third-party integrations, network
conditions, and data interactions. This simulation flags potential issues that may arise in production environments in real-time.

Moreover, GenAI explores these edge cases by generating test scenarios with extreme data inputs, rare user behaviors, or unexpected
system failures, uncovering hidden bugs or vulnerabilities. It also tests dependencies and integrations, simulating API calls, database
transactions, and communication protocols to verify reliability and compatibility with external systems.

Additionally, GenAI conducts performance and scalability testing by simulating heavy loads and concurrent user activities, stressing system
resources, and identifying performance bottlenecks or scalability issues.

3
3
Real-life applications of GenAI in
Healthcare software testing
Implementing GenAI into the STLC can revolutionize healthcare software testing across all user touchpoints,
be it in Device and MedTech software, Payer and Provider applications, or Life Sciences workflows.

4
Figure 1 shares a comprehensive overview of GenAI applications across healthcare software testing.

Figure 1: Overview of GenAI implementation in healthcare software testing life cycle

3
5
CitiusTech’s approach to QA solution
leveraging GenAI
Let’s consider the following scenario where QA team needs to perform the verification on the cardiology workflow.

Figure 2: General cardiology workflow

A general cardiology workflow is depicted in figure 2. A resident/fellow/referring physician/radiologist, who specializes in cardiology,
accesses the worklist through the EMR login page and identifies the priority cases to be addressed. Once a patient’s case study from
the worklist is selected, the relevant images and reports will be displayed on the workstation, where the physician/radiologist can
analyze the images and perform annotations. In the meantime, updates can be reconciled to the loaded study in the form of non-
DICOM information such as ECG. Cardiology measurements are performed by the healthcare professional, and a consolidated pdf
report can be generated. This report can be distributed for billing and further study purposes.

3
6
To verify the cardiology workflow, the QA professional will require to generate test scenarios, test cases, relevant test data,
automated test cases and execute them.

As illustrated in figure 3 below, GenAI can be leveraged to identify the test cases, generate synthetic data, and automate the
test cases. Let's take a closer look to understand the use cases of GenAI in QA, with reference to, a cardiology workflow and
how it can be implemented.

Figure 3: CitiusTech’s GenAI testing solution workflow

3
7
Test case generation Synthetic data generation Generating test cases for QA
As illustrated in figure 3, GenAI can GenAI could generate synthetic test automation snippet
readily analyze data across user data by mimicking the available data generation
touchpoints (highlighted in the cardiology in text, excel or data base format. For
The test cases generated using
workflow) using the ‘Document instance, it can help with study load
GenAI model can be used to train
Repository for Test Case Generation’. testing or cardiology measurements
the model again and generate the
For instance, GenAI can consume by generating synthetic patient data
automation script. This automation
traceability matrix, design documents, and continuing with the cardiology
snippet helps QA teams to execute
business requirement documents (BRD), workflow verification.
the process faster.
user manuals etc., to generate test
cases related to study loading scenarios,
cardiology measurements, and workflow
completion in standard format.

The user interface in the framework allows easy access to generated test cases and automated snippets. Thus, GenAI opens a plethora
of possibilities in QA across healthcare workflows, from reducing data dependency to improving processes. With test cases, synthetic
data generation, and automated execution, QA team can save manual efforts and time. However, enterprises must approach with
caution to ensure security and compliance is maintained.

3
8
The first step of GenAI in the QA journey:
Overcoming patient safety and privacy challenges
Before implementing GenAI for healthcare software QA, it's important to scrutinize potential patient safety and privacy challenges.
Here's how careful consideration can mitigate these concerns responsibly:

§ Ensuring transparency and explainability:

Healthcare enterprises can ensure GenAI-generated test scenarios are
transparent and understandable. This can be realized by explaining the
rationale behind each test case and allowing human testers to assess its
purpose and potential impact.

§ Protecting patient data:

To safeguard patient data, organizations can utilize privacy-preserving
techniques for generating diverse test data. QA teams can implement
privacy and security measures in the design phase, ensuring that patient
data is anonymized, and synthetic data adheres to privacy regulations.

§ Mitigating bias:
To tackle AI bias, QA team must continuously monitor and address
potential biases in AI algorithms used for test generation by employing
diverse training datasets to represent the full spectrum of patients and
healthcare scenarios. They can also utilize fairness metrics to evaluate and
mitigate bias in generated test cases, ensuring unbiased test coverage.

9
§ Maintaining optimal test coverage:
To enhance test coverage, QA teams must regularly evaluate
GenAI-generated test suites and supplement them with
human-designed test cases to ensure comprehensive coverage.
Deploying a blend of GenAI-generated and human-designed test
cases helps leverage both approaches' strengths for optimal test
coverage.

§ Enacting re-testing practices:

For agile application release, QA professionals must reimagine
proactive regression testing to minimize time and resources for
training. This can be followed by implementing robust change
impact analysis to identify areas of the software that require
regression testing after code changes. The approach helps focus
testing efforts on relevant sections and minimizes unnecessary
testing. QA teams can also apply AI-powered prioritization
techniques to prioritize regression testing efforts based on the
potential impact of changes on critical functionalities, ensuring
efficient resource usage.

10
Conclusion
The soaring medical software development market – accounting for $579.4 billion in revenue in 2023[5] – underscores the criticality of
high-quality healthcare software. This reflects the need for full-proof quality assurance in healthcare and the potential of GenAI in
transforming the QA process. However, enterprises need to approach this with caution.

Embracing GenAI should be about augmenting quality assurance rather than replacing manual testing. Enterprises can make the process
more proactive and agile by infusing human judgment and critical thinking into GenAI-governed quality assurance. This is achievable
across three stages: understanding their vision and scope for deploying GenAI, upskilling QA teams, and creating a customized strategy
for their workflows.

CitiusTech specializes in combining human expertise with GenAI to future-proof healthcare software. We draw from our deep domain
expertise to responsibly transform quality assurance tailored to healthcare workflow. We focus on delivering reliable, secure, and high-
quality healthcare solutions that benefit patients, and the entire healthcare ecosystem.

It’s time to embrace the next in healthcare software testing with GenAI.

References:
1. Global Medical Software Industry, Share, Analysis, Trends and Forecast 2022 – 2030 (acumenresearchandconsulting.com)

2. By 2028, GenAI-based Tools will be Capable of Writing 70% of Software Tests (idc.com)

3. The Impacts of Test Automation on Software's Cost, Quality and Time to Market (researchgate.net)

4. Top Trends Shaping the Future of Data Science and Machine Learning (gartner.com)

5. An Insider View of MedTech Software Development Trends (forbes.com)

11
Authored By:

Gaurav Shrimal
AVP, QA & Validation Practice, CitiusTech

Akash Gupta,
Sr. Technical Specialist, QA & Validation Practice, CitiusTech

Vamsi Vasireddy,
Tech Lead, QA & Validation Practice, CitiusTech

Kavita Jagtap,
Manager, QA & Validation Practice, CitiusTech

To learn more about this contact us at

[email protected]

visit our website

www.citiustech.com