The sheer volume of data generated by artificial intelligence can be a deterrent to adoption, yet harnessing AI fully can overcome that problem and benefit complex areas of healthcare such as organ transplantation, explains Tina Liedtky.

A key part of determining the success of stem cell and solid organ transplantation is human leukocyte antigen (HLA) typing, which helps predict how strongly a recipient’s immune system will recognise donor tissue as “self” or “foreign”. HLA and transplant immunology laboratories perform critical diagnostic testing to aid clinicians in selecting compatible donors, minimising immune rejection, improving graft survival and reducing complications such as graft versus host disease.

The integration of new technologies, such as AI, could improve HLA lab workflows and help laboratorians better manage and interpret complex data sets. While AI is gaining ground in laboratory settings across industries, adoption in HLA labs is slow due to several reasons including the ability to interpret the sheer volume of data that AI systems generate. That same technology, however, may offer a practical path to helping address it.

Transplant care requires timely and reliable data interpretation, and when clinicians are tasked with reviewing large volumes of data, delivering care can become more complex. AI has the potential to help laboratory professionals analyse and sort through data, flag trends and atypical results and, ultimately, support transplant teams in identifying information that may warrant further clinical evaluation.

While AI has strong potential in HLA labs, its adoption cannot be rushed in this setting because every model, workflow, step and output must be validated, documented and appropriately assessed, monitored and implemented in accordance with applicable regulatory requirements, institutional policies and laboratory quality standards before use in patient-care decision-making.

Ironically, this very challenge that is slowing AI adoption in the transplant lab, data interpretation, is an area where AI technologies may provide meaningful support. Labs that embrace the artificial intelligence technology with scientific rigour and validation may be best positioned to support improved clinical decision-making workflows.

Interpretation bottleneck in the transplant lab

HLA labs work with large amounts of complex data, so the potential to leverage technologies like AI to support faster analysis is significant. In these labs, there are various tests that must be conducted to determine a patient’s suitability for transplantation and to help assess a patient’s condition post-transplant. One of the most common tests performed in the transplant lab, HLA typing, requires labs to compare donor and patient profiles at the genetic level, detect antibody patterns and assess compatibility, all within tight timelines and often with high testing volumes.

While AI is gaining ground in laboratory settings across industries, adoption in HLA labs is slow due to several reasons including the ability to interpret the sheer volume of data that AI systems generate

Manual review of these layered datasets may contribute to bottlenecks that can affect workflow efficiency and care coordination. This is especially true in labs that handle high volumes of preand post-transplant testing data. AI models may complement traditional approaches to assessing graft rejection by supporting consistency in data interpretation and review processes. Additionally, by improving data interpretation and decisionsupport efforts, AI may help transplant teams make more informed organ utilisation decisions.

Where AI adds most value in the lab

There are many areas within the lab that AI can enhance to support more streamlined workflows, the biggest of which being data interpretation.

When transplant teams need to analyse multiple candidates, AI can flag off-trend results within patient data that might not be obvious in a primary review or single snapshot, which could help support additional clinical review during the transplantation process.

Depending on the AI model and its intended use, it may also assist in identifying patterns beyond traditional HLA matching approaches, support identification of potential changes in graft status and analyse trends in post-transplant biomarkers.

Additionally, AI has the potential to flag atypical outputs or irregular results and support longitudinal review of clinical and biomarker data that may be relevant to treatment management decisions, while helping labs maintain adherence to established protocols and procedures.

The critical need for scientist-led, AI-assisted decisions

Data interpreted by AI models can support the critical decisions that lab professionals need to make every day. However, AI needs to be able to explain to lab professionals why it flagged the result to help ensure outputs are appropriately reviewed and validated. To implement and use the technology responsibly, scientists must independently review and validate AI-generated outputs in accordance with laboratory procedures and applicable requirements. AI used in the lab should also be designed to minimise bias across factors such as age, race and gender. To manage this risk, regular monitoring of outcomes and validation of performance across diverse patient populations may be important considerations for laboratories implementing AI-enabled tools.

Building a culture around AI that prioritises transparency and scientific rigour may help support responsible AI adoption. Given the transplant lab's critical role in informing patient care decisions, that culture can also help ensure clinicians feel confident and appropriately supported when using the technology.

Ultimately, AI may help support more individualised clinical review and treatment planning discussions, no matter what stage of the transplant journey they are in, led by more comprehensive data interpretation.

• The views and opinions expressed in this article are those of the author and do not constitute medical advice, clinical recommendations, or regulatory guidance. References to artificial intelligence (AI) describe potential applications and emerging technologies in laboratory medicine. Any AI-enabled tools used in clinical or laboratory settings should be appropriately validated and implemented in accordance with applicable regulatory requirements, institutional policies and professional standards. Clinical decisions should remain the responsibility of qualified healthcare professionals.

• Tina Liedtky is president, transplant diagnostics at Thermo Fisher Scientific