News Stay informed about the latest enterprise technology news and product updates.

Telepresence recording gets smarter with searchable video, auto-tags

Who wants to sit through a 90-minute telepresence recording when only a fraction of it is relevant? In response, vendors are adding ways to search and tag content.

When users can't attend a meeting in person, telepresence makes them feel as though they're still in the room. But if users can’t make it to the meeting at all -- due to a scheduling conflict, significant time zone difference or because all the telepresence suites are booked -- they are often out of luck.

Anybody can capture all the video and audio in the world, but it's not going to be truly useful unless [the system] can auto-tag it, chapter it and [enable users] to view it based on certain keywords.

Dr. Neal Patel
Chief Medical Informatics Officer, Children's Hospital at Vanderbilt

Vendors tried to ease the scheduling crunch by adding recording features to their video conferencing and telepresence equipment, but that created other conundrums. Early servers and file storage systems for recorded video sessions were not intuitive, so users would struggle to find the right recording. Or they had no way of knowing that the final 15 minutes of a one-hour meeting was the only part they really needed to watch.  Now vendors are adding video content management systems to their products to make recorded video easier to find and use.

"Anybody can capture all the video and audio in the world, but it's not going to be truly useful unless [the system] can auto-tag it, chapter it and [enable users] to view it based on certain keywords," said Dr. Neal Patel, a critical care pediatrician and chief medical informatics officer (CMIO) at Children's Hospital at Vanderbilt, the teaching hospital for Vanderbilt University in Nashville, Tenn. "Nobody's going to talk like a robot and nobody's going to push a button to say, 'I'm talking about meds' or 'I'm talking about physical exams.'"

Vendors add 'social analytics' to video conference, telepresence recordings

Unified communications (UC) pros are looking for vendors to help them "make video [into] more usable content within an organization that can be mined and leveraged like other forms of content," said Zeus Kerravala, senior vice president and distinguished research fellow at Yankee Group.

Polycom recently announced its acquisition of Accordent Technologies, a video content management and delivery vendor, which will replace Polycom's original equipment manufacturer (OEM) relationship with Qumu, another video content delivery vendor. Polycom will support customers using the Qumu product for the next five years, according to Jim Kruger, vice president of solutions product marketing at Polycom.

Polycom and Accordent each already has "a strong relationship" with Microsoft's Lync and SharePoint products, Kruger said. When Polycom completely merges Accordent's technology into its RSS 4000 server -- which handles storage, streaming and recording for video conferencing and telepresence sessions -- the result will be a deeper integration with Lync and SharePoint that makes it easier for users to share and access video conference or telepresence recordings, he said.

"What we've seen a lot of customers ask for is beyond video conferencing: A time shift for those that couldn't be available," said Steve Pattison, senior vice president of strategic alliances at Accordent. "They're looking for social analytics [and ways to] make it searchable, indexable and easily found within SharePoint."

Both executives declined to specify further on the roadmap except that a deeper integration between Polycom's RSS 4000 and Accordent's Media Management platform would be available in the second quarter of this year.

On the eve of the Enterprise Connect trade show late last month, Cisco announced upgrades to its TelePresence Content Server and MXE 3500 transcoding appliance to make recording telepresence sessions and sharing that content easier for end users.

The server's latest software version, 5.0, which is due out this month, integrates with Cisco's Show and Share video sharing application and Cisco Quad, its social networking application, enabling users to post and access telepresence recordings directly through either platform.

Cisco also integrated its analytics software, Pulse, into the MXE 3500 -- an update available in the second quarter of this year. The application enables the server to use speech recognition software to identify and tag words and speakers in the video, allowing users to search for specific keywords or sections of a telepresence recording.

Although Cisco executives framed the announcement around telepresence, the features will also be supposed on Tandberg endpoints.

If an engineers at a company without these technologies missed a telepresence team meeting about a new product and only wanted to learn about changes to the product specifications, they wouldn't have to watch colleagues talk for an hour about marketing or sales efforts before getting to the relevant content, said David Hsieh, vice president of marketing for emerging technology at Cisco. With these newly integrated features, the user could search for a keyword such as "product specs" or a manager's name, the software would scan the speech-to-text transcript for either and the video would automatically jump to that segment, Hsieh said.

"Historically, these have all been separate systems, and while you could have [integrated them on your own], it would've required an enormous amount of custom work," Hsieh said. "Now, it's more like Legos."

Opportunities, limitations for auto-tag and search in telepresence recordings

Like many hospitals across the country, IT staff at the Children's Hospital at Vanderbilt has been gradually augmenting the old-school style of handwritten doctor's notes with electronic medical records and digital imagery. Clinicians have also been using Cisco's mobile HD video conferencing cart developed specifically for healthcare deployments, the Cisco TelePresence System Clinical Presence to capture and store recordings of patient care. The recordings can be played back at 1080p or 720p on a telepresence screen or other video display.

Recording, storing and accessing raw footage is simple to deploy and useful for brief or one-time encounters, such as uploading a personalized video on post-surgical care to a patient's individual Web portal, said Patel, the hospital's CMIO. But the applications for raw recordings are limited, he added.

"If you want to see a video, for example, of me explaining which medications you want to take home and how to change your [bandage], that could still apply," Patel said. "But to eventually use videos as part of your medical record, [we] need to have the auto-tagging and search functions."

Patel has been testing Cisco's latest upgrades to the TelePresence Content Server and MXE 3500 for the past month by having clinicians use mock cases in the hospital's intensive care unit (ICU) to capture what the actual environment might be like.

Enabling clinicians or compliance officers to search for keywords in a patient's video file, such as "pneumonia" or "physical exam," would improve adoption, patient care and regulatory compliance demands, he said.

"If I want to hear about the physical exam I should be able to cut right to it [in the telepresence recording] without watching the whole thing," Patel said.

However, the reality of a hospital environment has caused some snags with the speech-to-text software. The ambient noise in the ICU has given Patel inconsistent results, in terms of the software's ability to transcribe and tag portions of the video accurately or fully. He would also like to see Cisco enable users to provide more metadata about the patient encounter -- outside of what is actually said in the recording -- and enable IT to integrate the server more tightly with patient databases.

Without those additions, clinicians "would have to sit there and say everything, like, 'OK, now I'm starting the physical exam,'" Patel said. A clinician may not address patients by their full names, which would require them to point the camera at the patient's chart to have patient's name recognized by the software, he said.

Both of those actions would be unnatural for users and subsequently present major barriers to adoption, Patel said.

"It's not that it'd save time," he said. "It'd actually get people to use it because what we find in healthcare is if [a technology] is not easy to use, then people won't [adopt] it."

Let us know what you think about the story; email: Jessica Scarpati, News Writer.

Dig Deeper on Unified Communications ROI

Start the conversation

Send me notifications when other members comment.

Please create a username to comment.