# Non-invasive pressure sensor could revolutionize how brain injuries are diagnosed

An episode of [Physics World Weekly Podcast](https://invisiblenetwork.ai/podcast/1400686733), 10 Apr 2025, 26 min.

Audio: https://media.blubrry.com/physicsworldweeklypodcast/content.blubrry.com/physicsworldweeklypodcast/PWWeekly10AprilCrainio.mp3

Panicos Kyriacou of UK-based Crainio is our podcast guest

## Transcript

[0:08] Hello, and welcome to the Physics World weekly
[0:11] podcast.
[0:12] Our guest in this episode is the chief
[0:15] scientist of a startup company that's developed a
[0:18] noninvasive
[0:20] way to measure the pressure
[0:22] inside the skull.
[0:24] If their optical sensor
[0:26] gets approval for routine clinical use,
[0:29] it could revolutionize
[0:31] how brain injuries are diagnosed
[0:33] and treated.
[0:35] Here is Physics World's Tammy Freeman
[0:38] in conversation
[0:40] with Panikos
[0:41] Kyriaku.
[0:50] Traumatic brain injury
[0:52] caused by a sudden jolt or impact to
[0:54] the head is a leading cause of death
[0:56] and disability.
[0:58] After such an injury, the most important indicator
[1:01] of how severe the injury is is intracranial
[1:03] pressure, which is the pressure inside the skull.
[1:06] But at the moment, the only way to
[1:08] assess this is by inserting a pressure sensor
[1:11] into the patient's brain.
[1:13] Aiming to eliminate the need for such an
[1:16] invasive procedure,
[1:17] UK based startup Cranio has developed a noninvasive
[1:21] way to measure intracranial pressure
[1:23] using a simple optical probe attached to the
[1:26] patient's forehead.
[1:28] I'm joined today by Panikas Kiriakou,
[1:31] Cranio's chief scientist,
[1:33] to find out more.
[1:34] Welcome to the podcast, Panikas.
[1:37] Oh, thank you. Thank you so much for
[1:38] having me.
[1:40] So the first thing, can you can you
[1:42] explain why traumatic brain injury is such an
[1:44] important clinical problem?
[1:46] Oh, goodness me. I mean,
[1:49] just to give you a little bit of
[1:50] a background,
[1:52] can you imagine that every three minutes in
[1:54] The UK,
[1:56] someone is admitted to a hospital with a
[1:58] head injury? So it's a it's a very,
[2:01] very
[2:02] common problem, and you can imagine people having
[2:04] accidents,
[2:05] whether it's road accidents, whether any impact on
[2:08] the head to their work environment,
[2:10] in sports,
[2:11] contacts.
[2:13] So it's pretty common people have a blow
[2:15] on the head. Now how bad it is,
[2:17] nobody knows until actually they reach the hospital.
[2:20] So the ambulance will pick them up. It
[2:22] would drive them to an A and E
[2:24] naturally,
[2:25] and there, the patients will start having an
[2:27] assessment.
[2:29] And there will be a point that somebody
[2:31] from the neurosurgical team will be asked to
[2:34] look at the patient.
[2:35] Is it a concussion? Is it more serious?
[2:38] And, at the time of impact
[2:41] to the time that the patient will reach
[2:43] the hospital,
[2:44] and he will receive an assessment by an
[2:46] expert,
[2:48] especially in neurocritical
[2:49] care or in a trauma unit. That is
[2:51] known as the golden hour, and nobody knows
[2:54] what's happening to the brain at that time.
[2:57] And,
[2:58] it it it could be it it could
[3:00] be in a in a disaster in a
[3:02] way because
[3:04] you don't know how best to manage the
[3:06] patient, whether this patient has a severe traumatic
[3:09] injury, so the intracranial pressure is rising in
[3:12] the head, or it's a mild, or it's
[3:14] just a concussion, or it's a medium traumatic
[3:17] brain injury.
[3:19] So for many, many years, and especially in
[3:21] the last few decades, people are trying to
[3:24] find solutions in medicine
[3:26] where we can assess
[3:27] disease,
[3:28] diagnose,
[3:29] screen,
[3:31] disease
[3:32] in
[3:33] faster,
[3:34] noninvasively
[3:35] at the point of injury.
[3:38] And in the in the context of traumatic
[3:40] brain injury, it's something that at the moment
[3:42] cannot be assessed,
[3:44] at the point of impact.
[3:47] And in many ways, if a patient has
[3:49] a severe traumatic brain injury,
[3:52] the neurosurgeons,
[3:53] they will have to assess
[3:55] how bad is their intracranial
[3:57] pressure.
[3:58] Is it really above the threshold that classifies
[4:01] them as severe? And in order to do
[4:04] that, they have to drill a hole in
[4:06] the head, literally.
[4:08] And they will place into the
[4:11] brain
[4:12] a sensor probe. It's an electrical probe.
[4:15] And this is really,
[4:17] I I I would say, one of the
[4:19] most
[4:20] non
[4:21] the most invasive
[4:22] non therapeutic procedures.
[4:25] And,
[4:26] you can obviously do this,
[4:30] to every patient that comes in with a
[4:32] blow on the head.
[4:33] So it's it's quite invasive. It has its
[4:35] risks. Firstly, it has to happen within a
[4:37] clinical unit,
[4:39] with expertise.
[4:41] It could be,
[4:42] there is a risk of hemorrhage.
[4:44] There is a risk of,
[4:46] infection.
[4:47] And, therefore, there is this sort of cry
[4:49] out to develop technologies that can we measure
[4:52] intracranial pressure
[4:54] more effectively,
[4:55] earlier,
[4:56] and in a noninvasive
[4:58] manner? And for some people for many years,
[5:00] that it was almost like a dream. They
[5:02] say, no. Impossible to do that. How can
[5:04] you access the brain and see if the
[5:07] pressure is rising in the brain, and you
[5:09] can do this by placing something on the
[5:11] forehead?
[5:12] Now
[5:13] coming to us in Cranio, obviously, Cranio
[5:16] was born in
[5:18] 02/2022
[5:20] out of research, came out of City, University
[5:23] of London, now City, Saint George's University of
[5:25] London,
[5:27] and,
[5:27] within the research center for biomedical engineering.
[5:30] So this goes back in 02/2016,
[5:35] where the National Institute of Health Research gave
[5:38] us our first grant to investigate
[5:41] the feasibility
[5:42] of having a noninvasive
[5:44] intracranial sensor
[5:46] based on light technologies.
[5:49] The grant funded by the NHS was very
[5:52] successful. We managed to develop a prototype technology.
[5:55] We secured the intellectual property because we felt
[5:58] this is quite impactful.
[5:59] It has a commercialization
[6:01] value.
[6:03] And through that first study we did at,
[6:06] at the time, we managed to conduct a
[6:09] feasibility study on
[6:11] TBI patients, traumatic brain injury patients
[6:14] at the Royal London Hospital at the neurocritical
[6:17] care unit. And the Royal London in the
[6:19] East Part Of London is the biggest trauma
[6:21] hospital in U UK.
[6:24] And that was the time back in 02/2021,
[6:28] before Cranio was born,
[6:31] that we found
[6:32] we we discovered the first sort of positive
[6:35] news
[6:36] that the optical signals coming back from the
[6:39] brain
[6:41] after we apply an optical sensor. So we
[6:43] put a sensor on the head,
[6:45] on the forehead.
[6:46] We shine light into the brain. And a
[6:49] lot for many years and decades, people are
[6:51] studying
[6:52] the effect of light when it penetrates into
[6:55] tissue,
[6:57] including
[6:57] visible or infrared light, near infrared light. In
[7:01] in our lab, in my lab, we've done
[7:03] extensive studies on light tissue interaction
[7:06] and what happens to the light when it
[7:09] goes into the brain, passes through the forehead,
[7:11] through the skull, into the brain
[7:14] at certain colors of light, like near infrared.
[7:17] The light will come back, and the information
[7:19] coming back,
[7:21] this optical signal, also known the photopletysmogram
[7:25] or the PBG,
[7:27] it could really tell us there is information
[7:30] which within the signal
[7:32] that it keeps secrets of what's happened to
[7:34] the physiology
[7:36] or the hemodynamics
[7:37] of the brain.
[7:38] As you can imagine, when the pressure in
[7:40] the brain is rising,
[7:43] the pressure keeps
[7:45] the the the brain is swelling up, but
[7:47] it cannot go anywhere because the scalp is
[7:50] like concrete. And, therefore, the arteries and the
[7:52] vessels, they are compressed in the brain by
[7:54] that pressure.
[7:56] And this PPG signal
[8:00] shows us the changes
[8:01] due to the compression of the vessels.
[8:04] And, therefore, the optical signals
[8:07] change
[8:08] or components on the optical signal or features
[8:11] on the optical signal
[8:13] change
[8:13] with,
[8:14] changes in intracranial pressure. And there, people would
[8:18] develop the algorithms to be able to interrogate
[8:21] that optical signal
[8:23] and then develop models,
[8:25] machine learning models as we discuss in a
[8:27] minute, how to estimate,
[8:30] intercranial pressure.
[8:32] What is it that the the light's actually
[8:33] measuring? You say that the vessels are compressed.
[8:35] Is it is it looking at the actual
[8:37] size of the vessels or the the blood
[8:39] flowing within? Or
[8:41] The photoplethysmograph
[8:42] by its nature and definition, it measures volumetric
[8:46] changes
[8:47] of blood during systole and diastole.
[8:50] So as the blood pulsating through the arteries,
[8:53] through the cardiac cycle,
[8:55] the PBG, this optical technique,
[8:58] is able to show the changes in volume
[9:01] of blood within the artery.
[9:03] So you can imagine if you have a
[9:04] viscoelastic
[9:05] artery
[9:06] that is,
[9:08] opening and closing through systole and diastole, the
[9:11] volume of blood changes.
[9:13] And this is mapped, it's captured by the
[9:15] PVG.
[9:16] Now if you have an artery that is
[9:18] compromised,
[9:19] it is pushed down because of the pressure
[9:22] in the brain, that viscoelastic
[9:24] property of the artery
[9:26] is is impacted,
[9:27] and that would impact the PPG.
[9:29] So the changes of the PPG
[9:32] due to the volumetric changes it's experiencing
[9:36] due to the compression of the vessels in
[9:38] the brain
[9:40] can tell us information about the ICP.
[9:43] And
[9:44] following the research,
[9:46] within the university,
[9:48] Crennio was created. It brought together a team
[9:51] of experts, a lot of passionate people
[9:54] to aid the further development
[9:58] and commercialization.
[10:00] What makes this happen?
[10:01] CTE, the university, licensed the technology to NLC.
[10:05] NLC
[10:06] is a health technology venture builder
[10:09] based,
[10:10] headquartered in The Netherlands,
[10:12] And they work very closely with universities,
[10:15] medical schools,
[10:17] taking ideas,
[10:19] innovations
[10:20] in a way from bench to patient.
[10:23] So they gave us really that momentum to
[10:26] form the company. So Cranio is a UK
[10:28] based company.
[10:30] And The U and Cranio from the first
[10:33] initial days was very proactive,
[10:36] comprising of a small team with, people from
[10:39] expertise
[10:40] in medical devices
[10:43] and optical sensors.
[10:45] And the team worked tirelessly, really, in the
[10:47] last few years to generate,
[10:50] funding,
[10:51] to be able to progress further with the
[10:53] development of the technology, the optical sensor, to
[10:56] bring it to a technology readiness level that
[10:58] is ready for further clinical trials.
[11:01] So,
[11:03] in 2023,
[11:04] Cranio was very successful with the an Innovate
[11:07] UK biomedical catalyst grant, which will enable enable
[11:11] the company to
[11:13] engage in a clinical feasibility study,
[11:16] optimize the technology,
[11:18] to a more,
[11:20] advanced probe,
[11:21] both the optical sensor, the optical front end,
[11:24] and the analog back end, the analog front
[11:27] end, the electronics,
[11:29] and develop further the algorithms.
[11:31] Then later on,
[11:34] the company was awarded another
[11:36] National Institute of Health Research grant
[11:39] to be able to,
[11:41] move the technology more into a validation study.
[11:45] So now we reach the stage where Crenio
[11:47] redeveloped the sensor. It looks amazing.
[11:51] It looks very commercial when you look at
[11:53] it if you're in the business of, you
[11:55] know, medical devices.
[11:57] The
[11:58] technology has received,
[12:00] approval by MHRA,
[12:02] the UK regulator,
[12:04] in order to proceed to the clinical studies.
[12:07] The clinical,
[12:09] studies now are on route.
[12:11] Ethical approvals have been secured.
[12:13] We are again partnering
[12:15] with the Royal London Hospital.
[12:18] We also more collaborators came on board. We
[12:21] have people from Cambridge,
[12:23] from the traumatic brain injury team are joining
[12:26] us
[12:28] to support the team.
[12:30] And I think we're expecting to enter
[12:32] clinical trials by the end of this month.
[12:36] This will be an opportunity
[12:38] to
[12:40] work with a new technology, the new pro,
[12:42] much more advanced probe, much more advanced electronics
[12:46] that would enable
[12:48] more detailed acquisition of signals from TBI patients.
[12:52] So these are patients that are admitted in
[12:54] urocritical
[12:55] trauma units.
[12:57] And all of them, they do have an
[12:59] invasive intracranial
[13:01] ICP bolt.
[13:02] Hence, they will allow us to compare Yeah.
[13:05] Both standard with our signals.
[13:07] The signals then, they will be collated, and
[13:10] they will be analyzed by Kranios data science
[13:13] team. This is the team that is utilizing
[13:17] machine learning algorithms,
[13:19] basically looking at changes on the PPG signal,
[13:22] extracting
[13:23] morphological
[13:24] features from the signal amongst other parameters,
[13:28] and then building different models in order to
[13:32] develop the technology further in order to predict
[13:35] dynamically,
[13:36] noninvasively,
[13:38] intracranial pressure
[13:40] in either absolute measurements
[13:43] or indicative of,
[13:45] high ICP, low ICP.
[13:48] So we're at this stage now.
[13:50] The
[13:51] national and international interest in this project, it
[13:54] has been overwhelming. You can imagine
[13:58] the interest, first of all, from the neurocritical
[14:00] care communities.
[14:02] So the vibrations
[14:04] of cranios,
[14:06] you know, successes,
[14:08] have disseminated
[14:10] through,
[14:11] and there is a very positive feedback from
[14:13] the neurocritical care community.
[14:15] There is an active interest of the progression
[14:18] of the technology.
[14:19] But interestingly, now we we get a lot
[14:22] of interest, not peripherally,
[14:24] but from communities where
[14:27] injury to the brain is significant.
[14:29] It could be from sports associations
[14:32] like the rugby association.
[14:35] People are interested in concussion.
[14:37] People are interested in other pathologies relating to
[14:40] intracranial pressure. So even though Cranios' primary focus
[14:45] is to deliver a technology
[14:47] for utilization
[14:49] in your critical care, You can imagine
[14:52] that this technology could be used in emergency
[14:54] care, in ambulances,
[14:57] in helicopters. They transfer patients.
[15:00] So it can find this place in many
[15:02] NHS,
[15:04] places and beyond. So it's a very, very
[15:07] exciting
[15:08] journey.
[15:09] It's going well, and,
[15:12] we we are we are very positive on
[15:14] the outcome. Yes. So, I mean, the the
[15:16] device is portable, so it could be used
[15:18] at at the site of an accident.
[15:21] Exactly. The device being, one, noninvasive.
[15:24] The sensor, it's just a, like, a a
[15:26] sticky plaster that it goes onto your forehead.
[15:30] The back end is a little box with
[15:31] all the electronics.
[15:33] Obviously,
[15:34] from the past few years, we work in
[15:36] this research environment.
[15:38] So the technology, it was, connected into a
[15:40] laptop computer. This is what researchers do in
[15:43] labs. But now we are sort of transferring
[15:46] the knowledge into,
[15:48] a graphical interface. So you will have, obviously,
[15:51] device with a monitor
[15:52] to be able to,
[15:54] see the signals,
[15:57] see the values of ICP in a portable
[15:59] way. Yes.
[16:01] Okay. And then you you mentioned that the
[16:03] signals are analyzed by machine learning algorithms. So
[16:06] did you have to develop your own proprietary
[16:08] software? It was this part of the the
[16:10] process?
[16:12] That is correct. I mean, machine learning applications
[16:15] in health care, they are exponentially
[16:17] and, I would say, cataclysmically
[16:19] grown over the years.
[16:21] It allows people to develop models, whether it's
[16:25] in biosignals
[16:26] or imaging, using images
[16:28] to be able to diagnose disease.
[16:31] So the team
[16:33] within the university and later on by cranial,
[16:35] fortify
[16:37] the data science part of the team. We
[16:39] brought experts that they have spent years working
[16:41] with
[16:43] predominantly photopredismograms
[16:45] and be able to develop bespoke
[16:48] machine learning algorithms
[16:50] of extracting information
[16:52] from this physiological
[16:53] optical signal, the PVG,
[16:56] and built in models to be able to
[16:57] predict intracranial pressure. So, yes, it's correct. The
[17:01] algorithms developed by Cranio is almost
[17:03] like gold dust to the company.
[17:05] Okay.
[17:06] And as you say, you've done some preliminary
[17:08] tests on patients, and you're starting this this
[17:11] sort of new clinical trial. What do you
[17:14] what do you hope to achieve with with
[17:16] the next lot of measurements? What are you
[17:18] aiming for? Yeah. Absolutely. I mean, very good
[17:20] question. Of course, the first study, it was
[17:22] done during,
[17:24] the time where the project was within the
[17:26] university.
[17:27] It was the first feasibility study on a
[17:30] probe or sensor technology that was built,
[17:33] within the university.
[17:35] Since then,
[17:37] we've learned
[17:38] more,
[17:40] how we can you know, lessons learned from
[17:42] the first clinical trial
[17:44] on on a group of 50 patients or
[17:47] so.
[17:49] And the second round now conducted and let
[17:52] is is gonna be led by Crenio
[17:54] with a more optimized probe.
[17:57] So after the first experiences,
[17:59] we've learned,
[18:01] the technical challenges we had, and we're trying
[18:04] to mitigate them with the new probe design,
[18:06] whether it's in the optics,
[18:08] the configuration of the probe, the physical appearance
[18:11] of the probe, how it's placed on the
[18:13] forklift.
[18:14] So those lessons were learned. So, hopefully, this
[18:17] round,
[18:18] we will, mitigate some of the challenges we
[18:21] had before in the technical aspect.
[18:23] Also, through the first study, we've learned
[18:26] the challenges or the problems
[18:29] with the acquisition of signals.
[18:31] So the type of patients, how long we
[18:33] should monitor.
[18:34] So now we can do the study more
[18:36] rigorously,
[18:38] longer,
[18:39] more supervised in order to acquire
[18:41] high quality signals.
[18:43] This time around, we will record
[18:45] more information from the patients.
[18:47] We will, re we will look at CT
[18:50] scans
[18:51] to see how the scalp density, the thickness
[18:54] of the skull, perhaps
[18:56] it will have an impact on the light
[18:58] between, you know, different ages, different sexes.
[19:01] We will,
[19:02] monitor and collect data,
[19:05] from the commercial medical monitors within neurocritical
[19:09] care in order to see the relation of
[19:11] ICP with other physiological
[19:13] data acquired in these patients. So we are
[19:16] expanding the acquisition of data
[19:19] and the knowledge pool about
[19:21] what happens when a patient's intracranial pressure pressure
[19:25] rises.
[19:26] What happens to the blood pressures? What happens
[19:28] to the other physiological measurements of the patient?
[19:31] So we're enriching the study
[19:33] with a more advanced technology,
[19:35] and, hence,
[19:37] and fingers crossed,
[19:39] this would lead to more accurate machine learning
[19:43] models in relation to capturing correctly,
[19:46] dynamically
[19:47] the changes in intracranial pressure.
[19:51] And you and you said in this, in
[19:52] these patients, they're all they've also got the
[19:54] invasive probe. So Indeed. Compare the results and
[19:58] sort of check the accuracy.
[20:00] It is fundamental
[20:01] to be able to compare
[20:03] our results and our experiences
[20:06] in changes in ICP
[20:08] with the gold standard. So we will have
[20:11] all the,
[20:13] in real time, the traces, the physiological sick
[20:16] the signal coming from the ICP sensor
[20:19] into our own acquisition system so we know
[20:22] exactly
[20:23] what's happening to the patient's ICP
[20:26] according to the gold standard in relation to
[20:29] our own ICP technology.
[20:31] Okay. And then, you know, presuming all the
[20:34] the next,
[20:35] trials are all successful, how far away is
[20:38] the system from sort of being used clinically
[20:40] as as a standard tool?
[20:43] Well, Kranio is very ambitious.
[20:46] For those being in the medical devices,
[20:48] game, it's a long game, usually, medical devices.
[20:54] But the trajectory,
[20:55] we're hoping that within
[20:58] the next couple of years,
[21:00] we will
[21:01] progress
[21:03] adequately
[21:04] in order to enable us to, you know,
[21:07] jump all the regulatory sort of COPES,
[21:11] c marking,
[21:13] all the standards that are necessary to launch
[21:16] the medical a medical device.
[21:18] So we are quite optimistic. We think within
[21:20] the next two, three years, we'll
[21:22] be in a very good position actually to
[21:24] get very, very close to very close to
[21:26] the market,
[21:27] all all be well, of course.
[21:30] And then do you predict will this It
[21:32] sounds like it'd be sort of relatively low
[21:34] cost
[21:35] option if it's just based on optics for
[21:38] for hospitals to use.
[21:40] Well,
[21:41] usually, this is questions that speaks to the
[21:43] CEO.
[21:46] But,
[21:48] I mean, you are right. I mean, the
[21:50] ambition here is to the primary motivation
[21:52] of Cranio
[21:54] is to create solutions for health care. I
[21:56] mean, developing a technology
[21:58] which we can help clinicians
[22:01] to diagnose
[22:03] traumatic brain injury effectively,
[22:05] faster,
[22:06] accurately,
[22:08] earlier.
[22:09] It can only yield in the
[22:12] better patient outcomes,
[22:17] improving
[22:18] the quality of life of those patients. We
[22:20] work closely with the charity.
[22:23] And,
[22:24] through those discussions,
[22:25] they're very impactful for us within Cranio to
[22:28] see how the patients and the carers
[22:31] are impacted with traumatic brain injury.
[22:34] So that one that's one of our primary
[22:37] motivators. This is what gives energy to Cranio,
[22:40] really. Of course, as a as a company,
[22:42] we're interested in being successful commercially as well,
[22:45] but the ambition here is to keep the
[22:48] cost, first of all, affordable. We live in
[22:50] a world that take medical technologies, they need
[22:53] to be
[22:54] affordable, not only for the western nations, but
[22:56] the nations that they cannot afford state of
[22:58] the art technologies.
[22:59] So, this is, again,
[23:02] one of the primaries within Crenio to be
[23:05] able to
[23:06] create a sensor technology
[23:08] that it could be used widely. And
[23:11] used widely because there is a massive need,
[23:14] but also used widely because it's affordable technology.
[23:17] In the commercial journey of cranial,
[23:20] obviously, The UK is a big market.
[23:23] NHS, for us, it will be a primary
[23:25] a primary market. Obviously, our ambitions, they are
[23:28] beyond.
[23:29] We want to reach, obviously, the European market
[23:32] and, obviously, The US market, which is very
[23:34] big.
[23:35] So on our target is, at some point
[23:38] after,
[23:39] you know, the c marking approval,
[23:42] is to
[23:43] tap into The
[23:45] US market.
[23:47] So we need to obviously pass through the
[23:49] pathways of the FDA
[23:51] approval in order to enter The US market.
[23:54] So in the sort of short, medium, and
[23:56] long term, Crenio
[23:58] inspires by 2030
[24:00] to be able to penetrate
[24:01] The US market.
[24:03] Hopefully, the hopefully, The UK market as close
[24:06] as 2027,
[24:08] '20 '8.
[24:09] So we're ambitious. The team is amazing.
[24:13] All very passionate about the technology.
[24:18] So, yeah, that's the sort of endgame
[24:22] with the company.
[24:23] It sounds like a really promising tool that
[24:26] could you know, it has a lot to
[24:27] offer, Hunter.
[24:29] So I wish the company lots of luck
[24:30] for the future. That sounds great. Thank you.
[24:33] Thank you. Excellent.
[24:35] Well, thank you very much for joining us
[24:37] today. Thank you. Very welcome. My pleasure. Thank
[24:39] you so much.
[24:48] That was Kranios' chief scientist,
[24:51] Panikos
[24:52] Kyriakou,
[24:53] in conversation with Physics World's Tammy Freeman.
[24:57] I'm afraid that's all the time we have
[24:59] for this week's podcast.
[25:01] Thanks to Pankajos
[25:02] and Tammy for a fascinating conversation.
[25:05] And a special thanks to our producer Fred
[25:08] Isles.
[25:09] We'll be back again next week. But in
[25:11] the meantime, do check out the latest episode
[25:14] of the Physics World Stories podcast.
[25:17] Host Andrew Glester is joined by three
[25:21] expert guests
[25:22] to explore the impact of artificial intelligence
[25:26] on discovery,
[25:27] research, and the future of physics.
[25:31] That episode is called
[25:33] AI and the future of physics,
[25:35] and you can find it on the Physics
[25:37] World website
[25:39] or at your favorite podcast provider.

Shows are listed from Apple Podcasts' charts and directory, and Podcast Index where it is used. Every show and episode belongs to its publisher and plays from the publisher's own feed; this site hosts none of them.