The sustainability of medicine through machine learning. Artificial Intelligence in healthcare is often associated almost exclusively with diagnosis and personalised therapy, almost exclusively with precision medicine: Large databases are used to create algorithms based on experience to help clinicians diagnose and treat their patients.
The technology is expensive and progress is slow. Large hospitals that have the financial and human resources to afford the process towards AI benefit, and a project developed in one hospital is hardly transferable to others.
The reality is that, today, precision medicine for everyone and everything is not sustainable.
AI is applied for one-off, pioneering cases, with great media impact, but does it bring big enough changes for us to consider adopting AI in all hospitals?
The answer will have a myriad of different comments and assessments that will always lead us to fail to achieve health equity in different centres and regions.
The adoption of AI is a marathon, in which we cannot leave anyone behind. If we address high-impact problems, providing solutions for better management of available resources, the social benefit is for all.
Amalfi’s proposal is the use of AI in healthcare management, to help the sustainability of the system. Addressing major problems such as the management of absenteeism, which in our country consumes a large amount of economic resources and can be managed with AI obtaining cost reduction and improvement of labour welfare.
The methodology of ARUM (Analysis of Human Resources), based on Machine Learning, is to learn the behavioural patterns of employee groups from historical data and predict future absenteeism to improve planning, avoid staff shortages and reduce the costs of hiring substitute staff.
On the clinical side, can AI go beyond individualised, patient-by-patient treatment?
Again, the answer is yes. AI can also encompass the analysis, grouping and segmentation of all patients, providing answers to large population groups and providing solutions to very high impact problems without the need to apply large investments beforehand.
In the ANIS platform, the algorithm groups the patients described in the health records of the medical history (diagnosis, treatment, medication, etc.) in such a way that the groups created provide new information that has not been discovered or analysed with classic procedures, as it makes it possible to see which comorbidities, associations, procedures and treatments differ from the rest of the groups that were similar a priori.
This technology allows an analysis beyond aggregates such as DRGs (Diagnosis Related Groups) and the resolution of clinical hypotheses that respond to protocol changes and improve the treatment of patients, achieving efficiency of available resources. By identifying high-risk and high-cost groups, it is possible to focus action on part of the patients with a pathology and achieve an impact on the overall results.
We deal with data from the entire population that each organisation must treat, thus allowing us to move from Real World Data (RWD) to Real World Evidence (RWE).
