Public and Patient Information

AI Cancer Diagnostics: Patient Information

We are working with our Patient and Public Involvement (PPI) group to enhance public understanding of the benefits of AI in cancer diagnostics. This page has been created to give patients and the public more information.

At Sonrai Analytics, we understand how important it is for patients to receive accurate and fast cancer test results. That's why we're developing a cutting-edge Artificial Intelligence (AI) cancer diagnostic tool to help pathologists and clinical scientists in their day-to-day work and, as a result, speed-up time to results.


Our tool, developed in collaboration with the Precision Medicine Centre at Queen's University Belfast, is based on AI, which means that computers can be trained to "learn" and make decisions like humans. It's also based on cloud-computing technology, a fast, cost-effective, and secure way to access computing resources and software applications. It will allow pathologists and clinical scientists to upload images of biopsy and resection samples and run them through our AI medical devices. This will enable faster and more accurate patient test results for MSI Status associated with Colorectal Cancer and PD-L1 score in Non-Small Cell Lung Cancer.


Traditional cancer testing can be slow, expensive, and sometimes inaccurate. Our AI-based solution aims to change that. Patients can access the necessary treatments and further testing sooner with faster results. In addition, our AI-based approach also seeks to reduce the costs of testing for the NHS, bringing potential savings of up to £3.2m per year.


Our goal is to provide CE-marked clinical assistive tools for pathologists and clinical scientists that will allow them to deliver results in 24-48 hours. The CE mark is a certification that shows the tool meets the safety, health, and environmental protection standards in the European Economic Area (EEA). This means that the tool has been thoroughly tested and meets high standards of quality and safety. We're excited to bring this innovative solution to patients and the NHS.

Sonrai under the spotlight

The Public and Patient Involvement Group (PPI) ask the experts

Our solution is being developed with the active involvement of the patient advocates to enhance public understanding of the benefits of AI in cancer diagnostics. In the videos below, Dr Sandra Irvine and Tim Kerr of the Public and Patient Involvement Group get to the bottom of what an AI cancer diagnostic tool means for patients, oncologists and pathologists.

What Is AI And How Can It Benefit Cancer Patients?


AI can mean different things to different people. In this video, Dr Sandra Irvine speaks to Dr Yasmine Makhlouf about what we mean by AI in healthcare.

Will AI Diagnose As Accurately As A Qualified Pathologist?


How does AI learn? And will it deskill Pathologists or help them? Dr Sandra Irvine speaks with Machine Learning Engineer James Leech about how AI can help increase accuracy.

Will The Introduction Of AI Improve Patient Outcomes?


Will AI improve patient outcomes, and how? Tim Kerr asks Principal Data Scientist Matt Lee where patients can see AI's benefits today and where we will be in the near future.

How Do I Know That My Information Is Stored Safely?


How is patient data protected? Head of Engineering Gerard Loughran speaks to Tim Kerr about how patient data is stored and moved securely and the importance of balancing accessibility and safety.

Additional Information for Patients

Additional Information

Click on the links below for more technical information on Sonrai's cancer diagnostic solution, support services close to you or more information on the roadmap for NHS AI Lab.

Answering Your Questions


Do you have any questions for our experts that are not addressed below? If so, we're happy to answer them! Just fill in the short form below, and our team will get back to you asap.

AI has the potential to significantly improve cancer care by providing more personalised and effective treatments, improving patient outcomes, and reducing healthcare costs.

Early detection
AI can analyse large amounts of medical data to identify patterns and detect cancer at an early stage, which can significantly increase the chances of successful treatment.

Precision medicine
AI can help doctors personalise treatment plans for cancer patients based on their genetic profile and medical history. This can help identify the most effective treatments for each patient and minimise the risk of side effects.

Drug discovery
AI can help speed up the process of drug discovery by analysing large datasets and identifying potential new drugs that can be tested in clinical trials.

Predictive Analytics
AI can analyse patient data to predict the likelihood of cancer recurrence or the effectiveness of a particular treatment. This can help doctors make more informed decisions about patient care.

Remote Monitoring
AI can help monitor cancer patients remotely, allowing doctors to track their progress and adjust treatment plans. This can be especially beneficial for patients who live in rural or remote areas.

Robots and AI are two different technologies that can be used in cancer diagnostics.

Robots are physical machines programmed to perform specific tasks, such as sample preparation, image acquisition, and analysis. Robots can automate various processes in cancer diagnostics, such as staining tissue samples, scanning slides, and identifying abnormal cells. The use of robots can help improve the speed, accuracy, and consistency of these processes and can also free up time for human pathologists to focus on more complex tasks.

Conversely, AI is a broad field of computer science that focuses on developing algorithms that can perform tasks that typically require human intelligence, such as recognizing patterns, making decisions, and learning from experience. In cancer diagnostics, AI algorithms can be trained to analyze data generated by robots or other imaging technologies, such as CT scans and MRIs. For example, machine learning algorithms can be trained to detect patterns in images of cancer cells and differentiate between healthy and cancerous cells. This can help pathologists make more accurate diagnoses and identify potential treatment options.

AI has shown promising results in diagnosing certain types of cancer, but it is not yet as accurate as a qualified pathologist in all cases. AI algorithms can analyze large amounts of medical data and detect patterns that may be difficult for human pathologists to identify, which can improve diagnostic accuracy and speed up the process.

Our AI systems learn by example; the system is given a patient slide, and the answer is provided by a pathologist, which sets the theoretical limit of how accurate the test can become. If the answers we give are only correct 95% of the time, then the test can only be correct 95% of the time. As part of our design and development, we will be performing clinical performance studies against the current gold standard method of analysis for those cancer types to ensure our devices are safe and meet appropriate performance standards.

However, AI cannot replace the expertise and experience of a qualified pathologist who has undergone extensive training and has years of experience in examining tissue samples. 

That being said, AI can be used in conjunction with human pathologists to enhance diagnostic accuracy and efficiency. For example, an AI algorithm may be used to pre-screen tissue samples and identify areas that require closer examination by a pathologist. This can help speed up the diagnostic process and improve accuracy.

Overall, while AI has the potential to improve cancer diagnosis, it is not yet a substitute for human expertise in this area.

AI has the potential to automate some of the time-consuming tasks that pathologists perform, such as screening slides for abnormalities, identifying cancerous cells, and tracking disease progression. This could free pathologists to focus on more complex cases and make their work more efficient.

Pathology is a complex field that requires extensive training and expertise, and pathologists play a critical role in cancer diagnosis and treatment. While AI may be able to automate some aspects of their work, it cannot replace the experience, judgment, and human insights that pathologists bring to their work.

Instead, AI is more likely to complement the work of pathologists by providing them with additional information and tools to enhance their diagnostic accuracy and efficiency. For example, AI algorithms could help pathologists identify abnormal cells or patterns in tissue samples, which the pathologist could further examine and confirm. This can improve diagnostic accuracy and speed up the process but not replace the pathologist's expertise.

AI is more likely to enhance their work and improve patient outcomes.

AI can reduce the risk of patient misdiagnosis by providing doctors and pathologists with additional information and tools to enhance diagnostic accuracy. AI algorithms can analyze large amounts of medical data and detect patterns that may be difficult for humans to identify, which can improve diagnostic accuracy and reduce the risk of misdiagnosis.

For example, in the case of cancer diagnosis, AI algorithms can analyze tissue samples and identify abnormal cells or patterns that human pathologists may miss. This can help reduce the risk of misdiagnosis and ensure that patients receive the most appropriate treatment.

In addition, AI can help doctors and pathologists make more informed decisions about patient care by providing them with personalized treatment recommendations based on the patient's genetic profile, medical history, and other factors. This can help reduce the risk of misdiagnosis and ensure that patients receive the most effective treatment.

That being said, AI is not infallible and can make mistakes or miss important details. Therefore, doctors and pathologists need to use AI to enhance their work and not rely solely on it for diagnostic decisions. Ultimately, the diagnosis and treatment of patients should be based on a combination of clinical judgment, medical expertise, and AI-supported data analysis.

As part of the design and development of medical devices, engagement of potential users is essential, they provide vital information about their current workflows and risks associated with the sample type and indicate where new technologies could be used in the workflow. All of this information is used to identify where appropriate quality checks are needed in the AI workflow; software design can implement process steps at which the user must confirm they have performed a quality check on the sample, the data going into the AI for analysis and result alongside the AI decision rationale.

AI can potentially improve patient outcomes by enhancing diagnosis, treatment, and care accuracy and efficiency. By analyzing large amounts of medical data and detecting patterns that may be difficult for humans to identify, AI can help doctors, and other healthcare professionals make more informed decisions about patient care.

For example, in the case of cancer treatment, AI algorithms can analyze a patient's genetic profile and medical history to identify the most effective treatment options based on their specific condition. This can lead to more personalized and effective treatments that can improve patient outcomes.

In addition, AI can help healthcare providers monitor patients more closely and detect potential problems early on, which can help prevent complications and improve patient outcomes. For example, AI-powered monitoring systems can alert healthcare providers to changes in a patient's vital signs or other health indicators, allowing for earlier intervention and treatment.

Overall, while AI is not a silver bullet and cannot replace the expertise and experience of healthcare providers, it has the potential to improve patient outcomes by providing doctors and other healthcare professionals with additional tools and information to enhance the accuracy and efficiency of diagnosis, treatment, and care.

Yes, AI can improve with increased patient data. The accuracy and effectiveness of AI algorithms depend on the amount and quality of the data used to train them. With more data, AI algorithms can identify more patterns and trends and make more accurate predictions.

AI algorithms can be trained on large amounts of patient data, including medical records, imaging data, and genomic data. This data can be used to develop algorithms that can identify patterns and correlations that are not immediately apparent to human experts.

As more patient data becomes available, AI algorithms can continue to improve and become more accurate. This can lead to more precise diagnoses, personalized treatment plans, and better patient outcomes.

However, it is important to note that the quality of the data used to train AI algorithms is critical to their effectiveness. The algorithms may produce inaccurate results or make incorrect predictions if the data is biased or incomplete. Therefore, ensuring that the data used to train AI algorithms is diverse, representative, and of high quality is important.

AI is already being used in various ways for the diagnosis, detection, and personalized treatment of different types of cancer. Machine learning algorithms have been developed to analyze medical images, such as CT scans, MRIs, and mammograms, to help detect cancer earlier and more accurately.

In addition, AI can also be used to analyze genetic data and medical records to help identify personalized treatments that may be most effective for individual patients. This can lead to more targeted and effective therapies that may improve patient outcomes.

While AI is still in the early stages of development and implementation for cancer diagnosis and treatment, there is significant promise for its use in the field. As more data becomes available and algorithms are refined, AI will likely become more common across different cancer types for diagnosis, detection, and personalized treatment. However, it's important to note that AI does not replace human medical expertise and that medical professionals should carefully evaluate and validate any AI-based diagnostic or treatment recommendations.

The NHS has been at the forefront of the AI revolution with the creation of the NHS AI Lab, with these tools and products part of the £140m AI in Health and Care Award programme, each receiving a share of over £50m. The Accelerated Access Collaborative manages the award in partnership with NHSX and the National Institute for Health Research.

The package also includes funding to support the research, development and testing of promising ideas which could be used in the NHS in future to help speed up diagnosis or improve care for a range of conditions, including sepsis, cancer and Parkinson’s.

The NHS is committed to becoming a world leader in AI and machine learning, aiming to reap the benefits ranging from faster and more personalised diagnosis to greater efficiency in screening services.

Protecting data is one of our most important goals for Sonrai. We follow an ethos called - Privacy by Design and Privacy by Default. That means that when we design software or systems, those are the standards we measure ourselves by. We have ISO standards that detail how organisations test and validate their systems.

In the Sonrai system, we do not collect data that is identifiable. Sonrai works with pseudonymised data; this means that your existing provider will have an electronic health record with a patient ID which will be used to link up to a random IDNR system. We use imaging records and their ID, which can be traced back to the system. It does not retain patient data, it focuses on the required parts, and then your consultant returns to their existing systems.

No. Our medical devices will only be authorised to be used explicitly for the clinical purpose it has been designed for, which is to assist with diagnosing MSI and PD-L1.

It's very low risk. This is something we consider as a medical device manufacturer. When we perform cyber security assessments, we examine risk in two ways; 

  1. What is the probability or likelihood?
  2. What would be the severity?

We then design controls that mitigate those risks, such as encrypting the data at rest and transit, which means when it's being stored and accessed. We also insert multiple authentication controls that deal with who can access data and how they confirm their identity. We also segment the data so that if data is lost - it would not be identifiable. 

Because the data is encrypted when moved, you need a key to unlock it - otherwise, it is scrambled. In addition, we have envisioned that data might not need to move in how we have designed our platform and portal. The idea is that multiple consultants could log in to access the secure data, which removes any risk of data being passed from one site to another.

Sonrai uses cloud computing, which still means a server is sitting in a location. In our case, the server sits in London in one of the world's most secure data centres, Amazon Web Services. In practice, this means that patient data is encrypted in a server that has physical protection - including armed guards; it's constantly being monitored to ensure that data is backed up while at the same time being accessible. At Sonrai, we protect data while making it available to the right people at the right time.