We live in the information age, where digital data rules our lives. There's no doubt that computerisation and electronic records have transformed healthcare services. But there is a downside. Capturing the data during consultations creates a barrier between doctors and their patients. At Vision, we're using machine learning to transform healthcare software. We want doctors to interact with their patients, not their computers.
It's not too long ago that artificial intelligence (AI) was science fiction. Computers became more capable as the technology evolved. Autonomous systems are now an integral part of our everyday existence.
Optical character recognition (OCR) allows computers to read printed text. Speech recognition is everywhere; we take Siri, Alexa and Cortana for granted. We even have self-driving cars. Yesterday's AI turns into today's common technology.
What is machine learning?
Machine learning has gained new momentum. It's a branch of AI where systems learn from data to make reliable, repeatable decisions. These systems need minimal human intervention or re-programming.
It's very likely that you already have regular contact with machine learning. Your email spam filter uses it to remove junk emails. It can do this for any spam email, regardless of whether it has seen the sender or email content before or not. Have you noticed how search engines predict what you're looking for as soon as you start typing? That's machine learning too.
What if we apply the same technology to healthcare software? If we can predict a clinician's next move, we can:
- speed up the laborious job of adding data to patient records
- improve accuracy and data quality
- help clinicians to interact with their patient instead of the computer
It’s the little things that make a difference. We know you hate too many mouse clicks. You shouldn't need to tell the software what you want to do. The software should be able to work that out.
The usual way to add information is to specify what you're about to enter, and then add the data. To add a blood pressure reading you'd click on a blood pressure field and then type in the reading.
Vision predicts the kind of information you're adding, as you type. It's a little like predictive text messaging, but for clinical information. You add observations, diagnoses or medication into Vision's single data entry point. If you enter 140/120, Vision recognises this is a blood pressure reading.
There are many other common observations that Vision understands including:
- Weight and height
- Peak flow
- Oxygen saturation
In real-world consultations, clinicians must record many pieces of information. Vision makes this simple, accurate and very fast. You can see how easy it is to record common observations in Vision in this short video clip.
From free text to clinical codes
It's best practice to code consultation notes, but it's more natural to type free text. We're training Vision to pick out clinical terms from free text. As the clinician types, the software identifies and adds coded data.
We're integrating this as a natural part of recording SOAP-based consultations. As the doctor enters free text, Vision suggests appropriate codes.
Synchronise repeat medications
Most GP practices have a large number of patients with more than one long-term condition. These patients can have many prescription items. When a patient's repeat dates vary, practice staff must prepare their prescriptions several times a month, wasting precious time. It frustrates patients too. They have to keep an eye on their medicines to make sure they don't run out of anything. Organising prescriptions can be stressful too, especially for elderly patients.
The challenge is that it can take a long time to work out the dosages that will bring the prescription dates in line.
We have taught Vision to understand every possible dosage combination, including spelling errors. With two mouse clicks, the software synchronises the patient’s prescriptions. This saves valuable clinical time and reduces how often patients visit their practice.
There are other ways we can reduce the amount of computer time to enable quicker patient consultations. Doctors often need to look at a particular thread of their patient's medical history. It can be difficult to scan through the record and setting up filters can be cumbersome.
We're making this straightforward. Vision can automatically filter the information shown, so it is relevant to the information you're adding. If you're typing in blood pressure data, how great would it be to see all the previous readings, along with other relevant information such as medications and results like renal function? That's just how Vision works.
At a recent primary care conference, Vision's Chief Medical Officer, Dr Jonathan Behr, showed the power of intelligent healthcare software. Jon spoke the following dialogue between a patient and their doctor.
Patient: "Dr, I’ve found a lump in my breast."
Doctor: "I’m going to refer to you the breast clinic.”
Vision updated the consultation notes and created the urgent cancer referral at the breast clinic. It converted the Doctor's natural conversational speech into clinical codes. He never touched the computer. We’re building intelligent software so you can concentrate on your patients. Adding voice control is the logical final step.
The new Vision software will revolutionise how the NHS works.