Clinicians have traditionally spent significant time performing basic tasks that would not constitute working “at the top of their license” and where their involvement does not add value. There is also an acknowledgement that clinicians are already busy and that adding to their workload is unlikely to result in adherence to system improvements (Foley and Fairmichael 2015). This has resulted in significant interest in the automation of tasks. The process of automation has been underway in healthcare for many decades, driven by improvements in machinery and communication systems. Much of this would not constitute a Learning Healthcare System, but routine data and analytics offer the potential for automation to move closer to core clinical processes.
At Geisinger Health System, certain aspects of routine management and preventative interventions for well patients have been automated. This has increased the likelihood of these activities occurring and has freed clinicians to focus more time on managing the minority of patients with multiple active chronic conditions. Within consultations, a new initiative called SuperNote, has been launched to improve the quality of notes by prepopulating clinic notes with existing data. An “intelligent automation” function has also been integrated with the EHR, to pre-prepare tasks and orders that are likely to be required, based on the patient’s problem list (Foley and Fairmichael 2015).
IBM are currently developing their “knowledge driven analytics” system, Watson, to leverage unstructured information using Natural Language Processing (NLP) techniques. They hope that this will allow at least partial automation of two tasks that have been central to the role of the clinician:
- Leveraging the growing body of medical literature:
- It is not possible for clinicians to keep up to date with all publications, so the goal is to move beyond simple article search capabilities and train Watson to perform deeper analytics on the text, to understand the entities and the relationships described there to leverage further insight for the individual patient.
- Using the unstructured information in the patient history:
- A lot of what is captured in electronic notes is unstructured. There can be a lot of significant information contained within this, such as the justification for a particular treatment choice or the connections between symptoms, diagnosis and lab tests. In large longitudinal clinical records, there can often be over 1,000 pages. The goal here is to use Watson to produce accurate and relevant summaries of patient records.
Aspects of these systems are currently being trialled at Memorial Sloan Kettering Cancer Centre and at MD Anderson (Foley and Fairmichael 2015).
- A lot of what is captured in electronic notes is unstructured. There can be a lot of significant information contained within this, such as the justification for a particular treatment choice or the connections between symptoms, diagnosis and lab tests. In large longitudinal clinical records, there can often be over 1,000 pages. The goal here is to use Watson to produce accurate and relevant summaries of patient records.
There is significant scope for further traditional automation and for knowledge driven automation in healthcare, to save resources and to ensure that important activities are undertaken.
Evidence