AI & Precision Medicine

Precision medicine has long promised a world were health care intervention would be highly targeted, reliable and improve patient responses to varied treatment, mitigating risk associated with possible adverse reactions. Precision medicine is a conceptual framework in which clinicians, patients and other stakeholders work together to provide tailored treatments. Ultimately, this extends upon the patient-centred ethos that became the dominant paradigm in the early 2000s.

Precision medicine aims to bring various technologies together, expanding the arsenal of therapies available to practitioners, leading to highly targeted patient care. According to Orth et al. (Orth et al., 2019), advances in technology and increasing patient autonomy through the right to directly request tests and the use of lay specialists as substitutes have resulted in the introduction of precision medicine. 

Who benefits from precision medicine?

With the merging of so many technologies required to empower precision medicine, it is perhaps nieve to assume that the patient is the principal stakeholder; despite the current patient-centred paradigm. Lightbody (Lightbody et al., 2018), in summary, identified various stakeholders as part of the bioinformatics pipeline who are equally relevant within the broader context of precision medicine. Such stakeholders include; 

• patients 
• clinicians
• government agencies 
• insurance companies 
• Pharmaceutical companies
• standards agencies 
• NICE and various clinical guidelines agencies 

The challenges 

As one would expect, not all healthcare branches experience the early benefits of precision medicine despite the universal excitement surrounding the proposal (Salgado et al., 2019). It stands to reason that the growth of precision medicine within individual spheres of medicine and healthcare is mainly dependent on the following factors; 

• current treatment effectiveness and associated risks 
• the aggressiveness of disease following onset [specific to the patient]
• public health risks associated with the conditions specific to subgroup 
• public opinion 
• economic burden 

Luengo-Fernandez (Ward et al., 2013; Luengo-Fernandez, Leal and Gray, 2015) found a disproportionate funding allocation to cancer research and treatment (64%) over stroke (7%) and dementia (11%) between 2008 and 2018. These diseases are of interest as they account for 55% of all deaths in 2018.  Interestingly, Luengo-Fernadez (Luengo-Fernandez, Leal and Gray, 2015) and Ward et al. (Ward et al., 2013) show that interest shifted to stroke and dementia despite funding allocation and research. These areas remain underfunded compared with the burden associated with disease (dementia is the highest at £11m in social care costs) (Luengo-Fernandez, Leal and Gray, 2015). Thus, the implementation of services following funding trends and societal preferences are evident throughout healthcare (Salgado et al., 2019)(Ward et al., 2013; Luengo-Fernandez, Leal and Gray, 2015). 

Trust & AI 

Precision medicine is realised mainly in practice by using genomic medicine, AI decision & classification tools, and many other therapeutic devices based on AI and machine learning. It argued that such tools while only continue to advance so long as the public trust their use in practice. Genomic medicine, biobanks, genetic testing etc. have been the topic of heated debate for the past decade. 

‘socialising the genome’ project funded by genomics England, the welcome trust and the sanger institute aims to explore what people already understand about DNA and genomics (Filoche, Dew and Dowell, 2018).  Corresponding to the belief that for sufficient public acceptance of genomic medicine and the adoption of technologies in clinical practice, the public should be educated to use and understand technical, precise genomics terminology (Filoche, Dew and Dowell, 2018). While everyone does not share this view, it is acknowledged that the lack of public information creates a marked barrier to communication, creating an unhelpful power differential of expert versus others.

Ethical considerations & food for thought 

Precision medicine raises some interesting ethical questions such as:

1. data privacy: how secure is your genome 
2. ownership: is it your property, or are sequencing companies/biobanks allowed to see this information 
3. responsibility: how safe are such guarantees for the preservation of genetic information, what happens for loss of data. For example, many biobanks do not allude to what happens following DNA degradation. Instead, they focus on preserving stem cells and don’t offer any insight into the management of DNA quality. 

Similarly, who is accountable when a machine learning-based decision matrix decides upon the best treatment course for a patient? Does medical malpractice cover decisions when supported by machine learning tools?

References

Filoche, S., Dew, K. and Dowell, A. (2018) ‘Gnome medicine: what does genomic medicine mean to our patients and us?’, NZMJ, 131, p. 1473.

Lightbody, G. et al. (2018) ‘Review of applications of high-throughput sequencing in personalized medicine: barriers and facilitators of future progress in research and clinical application’, Briefings in Bioinformatics, pp. 1–17. doi: 10.1093/bib/bby051.

Luengo-Fernandez, R., Leal, J. and Gray, A. (2015) ‘UK research spend in 2008 and 2012: comparing stroke, cancer, coronary heart disease and dementia’, BMJ Open, 5(4), pp. e006648–e006648. doi: 10.1136/bmjopen-2014-006648.

Orth, M. et al. (2019) ‘Opinion: redefining the role of the physician in laboratory medicine in the context of emerging technologies, personalised medicine and patient autonomy (“4P medicine”)’, Journal of Clinical Pathology, 72(3), pp. 191–197. doi: 10.1136/jclinpath-2017-204734.

Salgado, R. et al. (2019) ‘Addressing the dichotomy between individual and societal approaches to personalised medicine in oncology’, European Journal of Cancer, 114, pp. 128–136. doi: 10.1016/j.ejca.2019.03.025.

Ward, D. et al. (2013) ‘Burden of disease, research funding and innovation in the UK: Do new health technologies reflect research inputs and need?’, Journal of Health Services Research & Policy, 18(1_suppl), pp. 7–13. doi: 10.1177/1355819613476015.