Artificial intelligence is conquering medicine - how algorithms could revolutionize palliative care

Wed, 2018 / 03 / 28

About the author

Ihre Ansprechpartnerin Dipl.-Kauffrau Heike Kielhorn-Schönermark

Dipl.-Kauffrau Heike Kielhorn-Schönermark
Founder and Managing Director

Fon: +49 511 64 68 14 – 0
Fax: +49 511 64 68 14 18

E-Mail More Contact Info

When it’s all about dying – an American study reveals that 60% of all deaths occur during acute hospitalization, where patients are usually intensively cared for in their last days of life. Although access to palliative care in the United States has improved in recent years (67% of all hospitals with 50 or more beds have palliative care in 2015, compared to 53% in 2008), there is still a lack of palliative care. According to the National Palliative Care Registry, less than half of those patients who need or desire palliative care actually receive it. Despite the expansion of palliative care structures, there is still a shortage. As the research team led by Anand Avati of Stanford University now explains, innovative technologies can play an important role in identifying patients who would benefit significantly from palliative care but fall through the grid of conventional classification models. Avati et al. address two known problems of palliative care referrals:

Physicians often do not even consider patients for palliative care. There are several reasons for this: too much optimism, time pressure or treatment inertia, that is, the tendency to remain in established treatment routines.

The "manual" identification of potential palliative candidates and the great lack of palliative care professionals make proactive classification expensive and time consuming. In addition, it is difficult to define criteria for deciding which patients could benefit from palliative care.

The Stanford team has developed an algorithm that makes recommendations for palliative care referrals based on the prediction of the mortality of a given patient within the next 12 months. Deep learning relieves palliative teams from tedious referral analyses, eliminating potential bias from human assessments, and enables objective recommendations for action, based on electronic health record data (EHR data). Machine learning is characterized by algorithms analyzing huge amounts of data and identifying common patterns and laws. The "learning effect" is that the system can then analyze and evaluate unknown data records of a similar type. The Stanford data model is currently being tested in a pilot project and will be adjusted based on the continuously expanding data situation and the learning experience of the system.

It is no longer just Facebook, Google, Zalando & Co., which use artificial intelligence or machine learning to optimize their chatbot functions in customer service systems, online purchase recommendations or advertising. Artificial intelligence is an indispensable part of the healthcare system, which becomes evident regarding the results of Anand Avati and the recently established marketplace for diagnostic imaging using artificial intelligence (similar to an app store).

As a long-experienced strategy consultancy in the healthcare sector, SKC follows the digital transformation of the healthcare industry with great interest and develops concepts for its clients to utilize digital potentials from an overall strategic perspective.

BY Heike Kielhorn-Schönermark, MBA, Founder and Managing Director, SKC Beratungsgesellschaft mbH and Beate Kasper, M.A. sociology, SKC Beratungsgesellschaft mbH

Sources:
Avati et al.: "Improving Palliative Care with Deep Learning"
Whatsnext: KI Radiologie
Whatsnext: New algorithms radiology revolution