GENOMICS CHART A Time allocation across BM development stages 10% 15% 20% 25% 5% 30%
Data collection and preparation
Model training Model evaluation and tuning
Development and integration
Research and experimentation Other supporting tasks
CHART B
Cost distribution across PHM development tasks 5%
20% 10% 25% 35%
Data acquisition and labelling Computational resources Software and tools Personnel Research and development
Fig 4. AI research resource for the PHM of HPO with biological modelling.
with public and private partners, to evaluate the safety and efficacy of BM classifications within advanced AI. Policy development, guided by the AISI, should ensure that AI governance prioritises patient safety and data security which develop through national initiatives and regulations.
• AIRR alignment: The AIRR seamlessly develop valid BM models with HPO classifications, actively engaging medical and scientific experts. Training programs developed by the AIRR should incorporate bias mitigation, performance metrics, and adherence to ethical guidelines and best practices. As illustrated in figure 4, the AIDRS HPO policy, HEMSS stewardship, and classifiers collaboratively align the AIRR on the PHM of HPO with BM deployments at scale.
n Recommendation C: International HPO classification ecosystem A standardised International HPO classification ecosystem provides global collaboration and data sharing. This should build upon extensive experiences across sectors that integrate science, technology, and innovation. n Science and technology classification: • Incorporate emerging scientific advancements, such as pangenome capabilities in multi-omics classifications, relevant to both UK and global research partners that benefit from classifications.
• Use advanced technologies like federated learning, hybrid AI, vision transformers, GPT-5, and quantum computing, supporting collaboration and knowledge sharing between the
42 UK and international partners.
• Implement advanced XAI and bias mitigation metrics to ensure equitable outcomes globally, with considerations for the UK’s diverse population to benefit from classifications.
n Classification proposal: • Standardise protocols for WGS germline and single-cell pre-eXams in BM, promoting interoperability with international efforts.
• Facilitate collaboration with biopharma and life-science companies internationally, to develop and implement precision-eXam intercepts.
• Promote the integration of practitioner hybrid AI and citizen choice in precision-eXam intercepts, considering ethical and practical implications.
• Establish clear international and national regulatory or accreditation pathways for the authorisation of generative AI in ‘X’.
• Conclude that the X in the Pre-eXam or the eXam defines its content, as such that a pre-eXam for a phenotypic trait may access multiple data sources. An eXam is neither restricted to biopharma intercepts with multi-omics gene therapy, oligonucleotides and proteins assessed by the AIDRS.
n Author’s position: • Emphasise the importance of robust assurance, building on previous work with a proposed stewardship for end- to-end workflows within ISO standard reviews and digital metrics for Gen AI classifications.
• Emphasise the importance of genomic medical services, biobanks and national imaging to establish research phases in the stewardship of HPO initiatives that can also contribute to global knowledge.
• Emphasise the importance of public-private partnerships, both nationally and internationally, for AI advancements in an era of quantum computing, federated learning, and Gen AI.
• Ensure that classifiers for Genome Pre-eXams and BM eXam intercepts, incorporating bias mitigation and XAI metrics, are implemented as a minimum classification standard nationally, and globally.
n Recommendation D: Continuous monitoring and evaluation
The HPO policy stewards evaluation mechanisms in an ecosystem of continuous improvement to ensure the long-term effectiveness of SDGs (Fig 5). n UK and US position on PHM ecosystem: • Prioritise the development of secure and safe digital twins with robust data protection measures, adhering to both national and international standards.
• Implement mechanisms for continuous improvement and accountability in AI development and implementation within the national healthcare ecosystem.
• Realise SDGs 3, 8, and 17 through promoting good health and wellbeing, advancing science and technology, and supporting collaboration between AIDRS, ICSs,
MAY 2025
WWW.PATHOLOGYINPRACTICE.COM 10% 5% 40% 30% 25% CHART D Focus of IRR development efforts
Specific disease areas Preventative care Early detection Resource optimisation Other areas
10%
20% 15%
30% 25% CHART C Innovation impact of emerging technologies in HPO
Advanced AI architectures
Improved data training techniques
Novel applications of AI
Federated learning platforms
Other emerging technologies
© copyright James Henry
Page 1 |
Page 2 |
Page 3 |
Page 4 |
Page 5 |
Page 6 |
Page 7 |
Page 8 |
Page 9 |
Page 10 |
Page 11 |
Page 12 |
Page 13 |
Page 14 |
Page 15 |
Page 16 |
Page 17 |
Page 18 |
Page 19 |
Page 20 |
Page 21 |
Page 22 |
Page 23 |
Page 24 |
Page 25 |
Page 26 |
Page 27 |
Page 28 |
Page 29 |
Page 30 |
Page 31 |
Page 32 |
Page 33 |
Page 34 |
Page 35 |
Page 36 |
Page 37 |
Page 38 |
Page 39 |
Page 40 |
Page 41 |
Page 42 |
Page 43 |
Page 44 |
Page 45 |
Page 46 |
Page 47 |
Page 48 |
Page 49 |
Page 50 |
Page 51 |
Page 52 |
Page 53 |
Page 54 |
Page 55 |
Page 56
