Health IT
Re: [MobileActive Discuss] Design Recommendations for IVR System in Kenya
Hi,
This is interesting enough that I also wanted to copy the health IT
community at GHDonline. Sorry for the mass emails.
Joaquin
---------- Forwarded message --------
On Jan 4, 2010, at 8:02 AM, Bryan Mc Alister wrote:
Good day,
I work in the Human Language Technologies (HLT) Research Group of the CSIRs Meraka Institute (South Africa), where we have recently completed a project that included a multi-lingual IVR system. The 'Lwazi' service, as it was named, essentially delivers telephone based information to community development workers in their specific regional languages within South
Africa.
The IVR system used to develop and provide the service comprises the following
- Asterisk (Software PBX)
- Kannel (SMS gateway)
- MobilIVR (Python convenience layer, developed by HLT Meraka):
MobilIVR is a convenience layer written in Python that sits on top of Asterisk and Kannel. It essentially allows application developers to easily create IVR applications in the Python programming language. It also provides communication models, which are easily accessible by app-developers. The
communication models are as follows: incoming call model; send/receive SMS; Call-back model where the user can leave an unanswered call, and have the service phone back again (This allows you to easily make the service free
for callers, especially in cases where it may be difficult to get a toll-free line)
** MobilIVR is essentially the latest version of the MobilEd software, just in case any of you are interested to know this **
We have been running the pilot service for nearly six months using one ISDN lines, and 4 phone numbers linked to that one line. In turn, we are making use of an ISDN-SIP gateway (provided by Patton).
For more information about the Lwazi service, you can take a look here:
http://www.meraka.org.za/lwazi/pilot.php
And here for the software that we used and developed ourselves:
http://www.meraka.org.za/lwazi/pilot.php
http://sourceforge.net/projects/lwazi/
http://sourceforge.net/projects/mobilivr/
This is a brief of our experiences, feel free to let me know if you would like any more details.
Regards,
Bryan McAlister,
Software developer
Meraka HLT, CSIR
---------- Forwarded message --------
On Wed, Dec 30, 2009 at 9:41 PM, Katrin Verclas wrote:
-- there is a bunch of us looking for some advice on IVR systems. See below. Any thoughts?
Happy New Year!
Katrin
---------- Forwarded message --------
From: Fouad Bajwa
Design Recommendations for IVR System in Kenya*
Dear Katrin and Mark,
I had the same questions a few months ago or a nationwide social enterprise medical centre deployment of 500 clinics that would be hooked up to both a non-profit hospital in Karachi, Pakistan and have an IVR system for primary or out door diagnosis at the the hospital that would host the IVR for the centre. Please include me on this thread as well because I want to see what choices are there?
The hospital is already using Open MRS, an open source HRMS and we were exploring possibilities of having the centres host video conferencing and IVR.
Regards.
--------------------------
Fouad Bajwa
Advisor & Researcher
ICT4D & Internet Governance
Member Multistakeholder Advisory Group (IGF)
Member Civil Society Internet Governance Caucus (IGC)
---------- Forwarded message --------
On Dec 28, 2009, at 10:49 AM, Mark Murphy wrote:
Hello,
I wanted to pass along some design thoughts we are having on an IVR system to be deployed in Nairobi. We are looking to benefit from previous experiences and to become acquainted with local experts.
The system will dispense information about local and nationaL resources for health, jobs, agriculture, education and finance. Most callers will access the system with their mobile using an advertised number. Using a series of menus they will locate the information of interest and be able to listen to it. Optionally they may request a follow up text message with address and phone number. The system will serve recordings in local languages.
Our requirements for the pilot are as follows:
* Must handle up to 5 simultaneous calls
* Must be as inexpensive to callers as possible
The proposed solution is to use a GSM gateway and Asterisk PBX server to receive and handle the incoming calls. The gateway would contain 5 SIMS and would be programmed to forward simultaneous calls to the next available SIM when multiple calls were being handled. A database server will be on the same LAN. The equipment would need to be housed in a secure place with good cell tower visibility to major operators.
There would be power backup and remote access to servers and gateway. The latter would require internet connectivity, possibly via a land connection.
We have a number of questions and would welcome the advice of members here.
* Would creating this set up a neutral location be preferable to hosting at an operator site. We're planning to find out more about what operators offer but are cautious about relying on a particular operator. Having ready access to the physical servers is preferred.
* If getting services from an operator in this regard is the best approach, which operator would people recommend?
* Would it be advisable to use SIMs from multiple operators with operator specific primary numbers to avoid inter-operator charges (we're not sure if this is still an issue in the Kenyan market)
* Would the high incoming traffic on the primary number of the gateway be considered a violation of standard usage agreements? We are fine with doing prepay or contract.
* We're wondering what lead time and expense we should allow for the internet connection of the LAN.
Are there factors we are overlooking?
http://www.MobileActive.org: A resource for activists using mobile technology worldwide
Jessica Haberer, MD, MS
Hi All,
It's great to see this thread.
I've been working with IVR in Uganda on an ART adherence project over the past six months or so. A company called Yo! Voice Solutions (www.yo.co.ug) in Kampala created the system using INTELLIVR and YBS software. It works really well from a technical stand point, but we've had some issues with the end user. We're targeting individual patients in very rural areas who are unfamiliar with this type of technology. We've learned a lot in terms of training needs, which I'd be happy to share if there is interest. Briefly, some key points are explaining how to use PINs, repeated training over time including tests of understanding, longer response windows, and visual explanations of the call algorithm.
Regards,
Jessica Haberer
Massachusetts General Hospital
Harvard Initiative for Global Health
2:04 PM, 5 Jan 2010 | Permalink
Jeff Rafter
I am very interested to hear the scope of these pilots/programs as they sound very exciting. In the case of Joaquin's post they mentioned 5 SIMs for 5 simultaneous calls. In my mind this seems very limiting for adherence if you consider a 10K patient facility where all (most) of the patients are taking drugs in the morning. What was the average time for an IVR interaction?
Thanks,
Jeff Rafter
8:38 AM, 6 Jan 2010 | Permalink
Aliya Walji
Hi Jessica,
I'm very interested to hear more about your project in Kenya. Currently, I'm working on a project where we would like to deliver maternal and child health information to pregnant parents in rural Ghana using both text messages and voice messages.
I'm speaking to Yo!Uganda shortly (since we have also worked with them for some of our work in Uganda) and we are also evaluating what technologies to use to build a voice system into our platform (called MoTeCH - Mobile Technology for Community Health).
However, one of the key areas I'd like to learn more about is how to design a voice system that will be usable by rural users who are often not literate or have experience with an IVR system. The issues you bring up are really interesting to me.
Could we talk more about your experiences and what you have learned (either off list or on this list)?
Thanks so much for sharing your experience in Kenya,
Aliya Walji
Technical Program Manager - ICT Innovation
Grameen Technology Center
+233540923923 • Skype: aliya.walji
6:06 AM, 14 Jan 2010 | Permalink
Jessica Haberer, MD, MS
Hi Jeff and Aliya,
expand commentSorry for the delay in replying.
The IVR interactions I've been using are short (1-2 minutes). The length of the call shouldn't really matter that much for capacity. A basic IVR board will hold 4 ports (i.e. 4 calls at once), but you can expand much larger than that for high call volume. The memory demands for the server are not really an issue either, although it's important to have duplicated backup systems in place.
My main project is in Uganda, although I'm working with another SMS project in Kenya. I'll check with that project lead to make sure he's open to sharing his experiences.
As for my project, please feel free to email me at to set up a call. Briefly, for the benefit of the community, I would emphasize the following areas:
1. Thorough training, building off the end users current knowledge base. Any use of PINs will require careful instructions, as most people are not familiar with them.
2. Test understanding, since a lot of people think and/or say they understand, but really don't.
3. Repeat training several times in the ...
10:02 PM, 14 Jan 2010 | Permalink