Computational pathology is becoming increasingly important in helping deliver precision medicine to a wider range of patients. Current developments span laboratory workflow optimisation, more informative inputs for molecular tumour boards and new capabilities for clinical trial design. The direction of travel is towards systems that connect previously separate data and processes, rather than tools that address a single task in isolation. Alongside established automation and analytics, newer approaches are emerging that are designed to operate across heterogeneous datasets and multiple modalities. This includes foundation models intended for rapid fine-tuning across downstream applications and agentic AI designed to move beyond content generation towards autonomous action within connected systems.
Integrating AI Across the Laboratory Pipeline
Computational algorithms are increasingly being deployed to optimise workflow and generate clinical impact by integrating data across the entire laboratory pipeline rather than focusing on single models. In pathology practice, Natural Language Processing can digitise request forms and generate medical reports, functioning as a virtual assistant within routine processes. Large language models are expected to reduce transcription errors and save time, although integration into Laboratory Management Systems remains a challenge.
Must Read: Securing AI Agents Against System-Level Threats
Within operational workflows, algorithms can support case triage, balance workload and enable real-time tracking and monitoring of laboratory processes. Automated quality control mechanisms can detect mismatches and flag errors. In dermatopathology, algorithms can estimate the likelihood of melanoma or carcinoma while prioritising urgent cases. Laboratories can automatically route melanocytic lesions to local experts, facilitating faster and more accurate decision-making.
Foundation models are reshaping pathology by offering pre-trained architectures that can be fine-tuned for multiple downstream tasks. Because they do not begin each task from scratch, they support scalability and robustness. Beyond generative AI, which creates content, agentic AI systems are designed to decide, act and interact with other systems. Such systems can draft medical reports autonomously and oversee laboratory operations, identifying bottlenecks and executing corrective actions. The transition from observation to action marks a significant evolution in how artificial intelligence may function within pathology services.
Multimodal Data for Precision Oncology Decisions
A multimodal approach is being advanced to deliver more specific information to molecular tumour boards and support precision medicine. Integrating multiple data streams into precision oncology programmes can enable the identification of well-defined patient populations with heightened sensitivity to specific therapeutic interventions. Data science underpins efforts to standardise clinical data, patient-reported outcomes, multiomics, spatial omics, digital pathology and radiomics.
Standardising and combining data streams can make complex information easier to interpret across teams. Data retention is intended to enrich subsequent decision-making processes. A significant step has been the creation of the Swiss Personalised Oncology network across major hospitals and universities. Structured data and Natural Language Processing are used to extract information from diverse medical record systems, enabling longitudinal follow-up of patients. Time-bound data can be analysed with process mining technology to understand how patients move across treatment lines.
The spatial component of multiomics is considered particularly relevant. Spatial biomarkers have been identified as predictive of benefits of immunotherapy in melanoma, lung and other cancer types sensitive to immunotherapy. In response, a spatial workflow has been developed that can be applied to digitalised or physical slides. A fully digitalised approach has led to the development of a predictive image-based digital pathology biomarker for immuno-oncology therapy response in metastatic melanoma. In parallel, a national omics network in Switzerland has been established to recruit 300 patients, with treatment decisions based on next-generation sequencing programmes, spatial omics, digital pathology and drug screening. Tools are being built to enable interactive data visualisation and discussion during tumour board meetings, bringing clinicians and data scientists together around integrated datasets.
Digital Twins and AI in Clinical Trials
Beyond diagnostic workflows, digital technologies are expanding into new domains of precision medicine. Digital twins, which connect real-world processes with virtual representations, have progressed rapidly from conceptual development to practical implementation. Their full potential in precision medicine remains under exploration.
Computational pathology is also influencing clinical trial design and execution. Its applications include automated biomarker qualification, tumour microenvironment analysis, prognostic and predictive modelling and quality control. By embedding algorithmic analysis into trial processes, pathology data can contribute to more structured and reproducible evaluation of therapeutic interventions.
These developments reflect a broader transformation in pathology, where computational tools are integrated into diagnostic, operational and research environments. The convergence of multimodal data integration, spatial analytics, foundation models and agentic AI suggests a trajectory towards increasingly automated and data-driven systems capable of supporting both routine laboratory management and advanced oncology decision-making.
Advances in computational pathology are extending from workflow optimisation to multimodal precision oncology and clinical trial innovation. Algorithms are being embedded across laboratory systems to support triage, quality control and reporting, while foundation models and agentic AI introduce new capabilities for autonomous decision-making. Multimodal data integration and spatial biomarkers are strengthening the information base available to molecular tumour boards. At the same time, digital twins and computational methods are influencing clinical research. These developments illustrate how artificial intelligence is moving pathology from analytical insight towards coordinated action within integrated healthcare environments.
Source: Healthcare in Europe
Image Credit: iStock
