Name: Phil O'Sullivan on good clinical governance and AI in health
Uploaded: 2023-10-26T04:12:25.266Z
Duration: 212.27
Description: Phil O'Sullivan, Partner at Allens, offers a concise masterclass on responsible data governance in healthcare. Learn how to navigate challenges, develop clinical evidence, and balance governance for success in this dynamic industry.

Phil Edited.mp4

Phil Edited.vtt

Subtitles

{{00:00:06}}
You need to have continual

{{00:00:07}}
monitoring and validation.

{{00:00:09}}
You need to understand how

{{00:00:10}}
the data was tested before

{{00:00:12}}
it's used and means to have

{{00:00:13}}
it clinically validated and

{{00:00:15}}
ongoing testing in place.

{{00:00:36}}
So the key challenges

{{00:00:37}}
for an organisation is to

{{00:00:39}}
understand its use cases and

{{00:00:41}}
frame them up, understanding

{{00:00:43}}
the regulatory framework.

{{00:00:44}}
And start to develop

{{00:00:45}}
guardrails so that the

{{00:00:46}}
organisation can engage in

{{00:00:48}}
deliberate uses of data.

{{00:00:50}}
And also establish good

{{00:00:51}}
governance so that there's

{{00:00:52}}
a framework for people to

{{00:00:54}}
address internally how to

{{00:00:56}}
get non approved use cases

{{00:00:57}}
through the system and

{{00:00:58}}
process that you set up.

{{00:01:00}}
Good clinical governance

{{00:01:01}}
to us means cross

{{00:01:03}}
functional governance.

{{00:01:04}}
So you involve all the

{{00:01:05}}
subject matter experts,

{{00:01:06}}
whether it's technology,

{{00:01:07}}
clinical safety, legal.

{{00:01:10}}
And they can all come

{{00:01:10}}
together and develop the

{{00:01:13}}
path forward for these

{{00:01:14}}
deliberate intended use cases.

{{00:01:17}}
You need to have continual

{{00:01:18}}
monitoring and validation.

{{00:01:20}}
You need to understand how

{{00:01:21}}
the data was tested before

{{00:01:23}}
it's used and means to have

{{00:01:24}}
it clinically validated and

{{00:01:26}}
ongoing testing in place.

{{00:01:28}}
So there's lots of checks and

{{00:01:29}}
balances you need to set up.

{{00:01:31}}
All of that can come through

{{00:01:32}}
a good governance committee

{{00:01:33}}
where you steer those pre

{{00:01:35}}
approved use cases and

{{00:01:36}}
frameworks through the

{{00:01:37}}
guardrails you create.

{{00:01:39}}
So we've worked with an

{{00:01:40}}
organisation that uses

{{00:01:42}}
AI in radiology and in

{{00:01:45}}
ophthalmology and they've

{{00:01:47}}
been able to develop the

{{00:01:48}}
technology so that operates

{{00:01:49}}
as a QA tool over the top

{{00:01:51}}
of clinical decision making.

{{00:01:53}}
So it's not seeking to replace

{{00:01:55}}
clinical decision making,

{{00:01:56}}
but while it's been used

{{00:01:58}}
in quality assurance, it's

{{00:01:59}}
able to actually establish

{{00:02:01}}
its own evidence base and

{{00:02:03}}
it's able to be tested in

{{00:02:04}}
a controlled environment.

{{00:02:06}}
It gives feedback to practice

{{00:02:08}}
managers and to clinicians

{{00:02:09}}
when needed but it's not an

{{00:02:11}}
immediate full on use case.

{{00:02:13}}
There's a period where you

{{00:02:15}}
can adapt the technology,

{{00:02:16}}
understand what results it

{{00:02:18}}
puts out, train it to get to

{{00:02:20}}
the better outcome more often.

{{00:02:22}}
And so that's a good way

{{00:02:23}}
of effectively rolling out

{{00:02:24}}
a pilot project and then

{{00:02:26}}
allowing that to prove itself

{{00:02:28}}
over time before going on

{{00:02:29}}
for the full on project you

{{00:02:31}}
can engage with the tech.

{{00:02:32}}
So the balance is important.

{{00:02:34}}
You can't have a governance

{{00:02:35}}
framework and treat all new

{{00:02:37}}
activities under the same

{{00:02:38}}
standard because resources

{{00:02:39}}
are tight, attention is tight.

{{00:02:42}}
And so you need to develop

{{00:02:43}}
a process where you can

{{00:02:45}}
have a tiered system,

{{00:02:47}}
understand the higher risk

{{00:02:48}}
activities and maybe build

{{00:02:49}}
more governance around that.

{{00:02:51}}
And also to roll out projects

{{00:02:53}}
that might be lower risk

{{00:02:54}}
with a tolerance for some

{{00:02:55}}
failure and also learn from

{{00:02:57}}
those failures as you go.

{{00:02:58}}
So it's more important to

{{00:02:59}}
really assess what the risk

{{00:03:01}}
tolerance is at the highest

{{00:03:03}}
levels of the organization,

{{00:03:04}}
the executive, the board, and

{{00:03:06}}
try and establish those use

{{00:03:07}}
cases where you can roll out.

{{00:03:09}}
Easy wins, quick success,

{{00:03:11}}
and evaluate how they go.

{{00:03:12}}
And you might also build

{{00:03:13}}
more formal, sophisticated

{{00:03:15}}
governance structures

{{00:03:16}}
around the really

{{00:03:16}}
hard clinical decision

{{00:03:18}}
support type use cases.

{{00:03:21}}
And then pursue those

{{00:03:22}}
coming off the evidence

{{00:03:23}}
base and learnings from

{{00:03:24}}
your other projects.

Expert Insight

Phil O'Sullivan, Partner at Allens, offers a concise masterclass on responsible data governance in healthcare. Learn how to navigate challenges, develop clinical evidence, and balance governance for success in this dynamic industry.