The researchers analyzed the audio files for their quality as well as the accuracy of the CERT members recording a complete response. Every voice recording had some kind of noise that was caused by the electronics (mechanical sound) or background interference (human voices and other environmental noise). This made it quite cumbersome to extract the information from the recordings. In several occasions the researchers had to listen to the audio recording more than once to determine the answer the participant had provided.

Figure 1 shows that on average only 85% of the information pieces could be recovered. A piece of information would mean an answer to a question. There were ten questions total. Given that the intent is to use FF with voice for exchanging emergency information such as incident situational reports. Incomplete information may cause delays and inappropriate assignment of response resources. Typically an emergency coordinator at the incident management hub (in this case the Sarvodaya Hazard Information Hub) would have to call back the CERT member to recover those missing pieces. If the incidents were life threatening, then such uncertainties are intolerable. Furthermore, automatic the transformation of the speech to text would be impossible and unreliable.

The outcome of the training and survey are discussed in the workshop report.

Based on the RTBP findings, one of the policy recommendation was that each hospital should employ an assistant to digitize the health records. The incremental cost in hiring a new resource person does not affect the total cost of ownership. During this mS-cube project, the Wayamba Province Health Ministry hired hospital assistance. However, the recruitment was politically influenced, which resulted in hiring English illiterate persons that had no aptitude for digitizing health records.

To strategically introduce an electronic version of disease surveillance and notification system, in developing countries, would be to align the RTBP with the existing Sentinel Site Surveillance (S3) programs. The S3 data collection, inferencing, and dissemination can be enhanced with the use of mobile technology. This project developed a software requirement specifications for developing such a mobile software suite for supporting the S3 programs. The specifications are on three components required for disease surveillance and alerting: collection of outpatient, inpatient, and special disease investigation data.

The mS-Cube project report

This article was published in the May edition of “Searchlight South Asia,” a monthly newsletter by Intellecap that tracks trends in pro-poor development for the Rockefeller Foundation. The opinions expressed in this article are solely those of the authors and do not reflect the positions of the Rockefeller Foundation.

What we were more eager to hear was the defense on the claim that the “web is dead, long live the internet“. In defense – “No the web isn’t dead” with the success story pointing to the Wikipedia. Was the WWW supposed to simply HTTP posts, gets, and requests? Or were people supposed to conduct business over the web?

The Common Alerting Protocol (CAP) was developed by Art Botterell et (2006) with the WWW and the underlying Internet, as the basis for exchanging emergency messages but with an all-hazards all-media approach. The LIRNEasia research, mostly on alternative, none PC and Internet, technologies has been studying the CAP implementation challenges in the last-mile and challenges with introducing the protocols to frontline personnel in developing countries. The knowledge attained has been transferred into developing the Sahana Alerting Broker with the aim to develop a comprehensive publisher-subscriber tool for emergency managers.

Paper on RTBP CAP-Profile

WIS CAP Workshop Report

This report summaries the learnings from the CAP Implementers Workshop hosted by the World Meteorological Organization Information Systems, International Telecommunications Union – Standardization (Recommendation X.1303), and the Organization for the Advancement of Systems and Information Standardization Technical Committee. The fourth CAP meeting (06-07 April 2011) was held at the WMO office in Geneva, Switzerland.

publishers, subscribers, and brokers (image taken from Bigham's ICCIA slides)

Those who know graphs theory are familiar with the “four color theorem“; an example being the world map (relaxed as a planar graph) can be colored with a minimum of four colors such that two countries sharing a border do not share the same color. Researchers at Queen Mary University of London use this theorem to color code cellular network base stations. The base stations are in abstract sense regarded as message Brokers (also termed as “Publisher Subscriber Message Oriented Middleware” – PSMOM) that channel the published message (SMS, Email, Voice packet, data packets, etc) to the Subscriber (or message recipient). Sometimes a subscriber and a publisher can be directly linked through a single broker or they may be linked through several intermediary brokers.

The role of the Sahana Alerting Broker, essentially, is similar to that of a cellular base station; where decision-maker or decision-system published risk information is disseminated to the subscribers of the response systems in the form of public warnings or restricted and private alerts (also known as closed user group alerts typically applicable to first responders). There are two downstream message distribution methods: Direct and Cascade; where direct alerting is a broker-to-person and the cascade method delivers the alerts to another broker (system); i.e. broker-to-broker. In addition, these broker, publisher, and subscriber network models would inherit one or a hybrid of the typical network topologies.

Sahana Alerting Broker subsystems (image taken from book "Biosurveillance methods and case studies", chapter 14)

Sahana Alerting Software for Real-Time Biosurveillance in India and Sri Lanka (soon available in IEEE Xplore) was the topic of my paper and presentation at the 1st International Conference on Computer and Information Applications 2010, held at the Tianjin Polytechnic University in China (Dec 03 – 05, 2010) – Click to view presentation: power point and video. It was nice to have spent some time with the keynote speaker Prof. John Bigham, Technical Advisor, School of Electronic Engineering and Computer Science, Queen Mary University of London, chatting about brokers at this event. Besides Porf. Gordon Gow, with whom I have been working on the Common Alerting Protocol (CAP) enabled Sahana Alerting Broker for the past several years, Prof. Bigham is the first person I came across who was thinking of Brokers in the same way; even though his work was more on the physical layer opposed to Sahana Alerting Broker being in the application/content layer. We both agree that there was very little research and literature available in the context of measuring Brokers for their robust-reliability, throughput associated performance, or other characteristics. This was a dilemma in working towards classifying early warning systems with the need to find ways to place a universal measure on brokers. Message brokers play an intermediary role between the decision subsystems and the response subsystems in the early warning chain of subsystems.

now back to the four color ant story … An ever present problem that pester those communications engineers is optimizing the quality of service; thus reducing characteristics like latencies (round trip time), jitter, etc (see LIRNEasia work on Broadband QoS). Another important aspect is the end-to-end resilience that comes along with crises where network nodes are broken and the are pipes blocked. You may have heard of emergency services (police, ambulance, fire-brigade, military, etc) using ad-hoc networks with multiple hops. During a crisis, the emergency services need to work within the constraints and require that the brokers (or routing agents) are smart enough to carry their traffic without disrupting their communication. The revenue generated through these priority occupants – emergency services – would be more than your’s or my personal phone call; hence, our call would get dropped in such a setting. Researchers at Queen Mary University of London choose to use a revenue penalty formula as the objective function to optimize the network routing protocols to ensure end-to-end resilience.

How does it work? Once the the base stations (or brokers) are colored then the Ant Colony Optimization (ACO) algorithm is applied to each clique of base stations with equal color to find the optimal paths within those cliques. The iterative ACO process is a technique typically used in such adaptive problems or combinatorial problem. In the broker resilience problem the best paths are those that have a lesser pheromone trail; i.e. pipes that are less congested (or blocked). The adaptive methods use semi smart antennas that cooperate in tilting or yawing to maximize the radio coverage and signal strength.

Computer scientist first developed the ACO algorithm as a solution to the typical Traveling Salesmen Problem (TSP). While the Ant algorithm is know to be slower than competing similar shortest path algorithms, but with less computational complexity, is good for tackling certain problems; such as finding the set of optimal paths for any combination of subscribers and publishers in a cellular network. These optimal strategies do not require the calculations to be executed frequently, perhaps once a day or less frequent as once a week.

The problems associated with Sahana Alerting Broker has not evolved to this level of path breaking combinatorial complexity. However, the complexities are more in CAP with the need to address multi natural language translation for quick relaying of alerts for rapid onset hazards that are constrained by very little time spans. Others challenges are more on the social and policy aspects of getting users to realize the importance and adapt such systems in developing countries. The Sahana Alerting Broker in the RTBP pilot had shown promising results to the extent the Public Health Inspectors in Sri Lanka have been frequently utilizing the system to share detected adverse public health events with targeted health officials and health workers. They also improvise the software to communicate other investigation information, otherwise would require a long bus ride for several ours to the health department to receive that information on paper, now is relayed electronically at near light speed on to their mobile phones!

Ratio between noisy and clean data for India and Sri Lanka

Fidelity of digitized data in the Real-Time Biosurveillance Program (RTBP) was not promising; especially with the personnel in Sri Lanka with no medical knowledge but technically capable were producing up to 45% noisy data (second stacked graph). On the contrary the medically trained but less fluent in mobile phone usage Indian nurses were less prone to producing noisy data. The Indian health workers had an incentive because the erroneous data would produce false alarms, and they would need to respond to these false alarms or it would portray a bad image of the health situation in their area; while the Sri Lanka data digitizing personnel had no incentive besides picking up a paycheck for the data entry work they did. The data was submitted through the mHealthSurvey mobile software that works on less expensive Java-enabled hand-helds.

High counts of fever cases in a single location on a single day

The RTBP envisions that hospital data is submitted each day; thus, the real-time expectations. However, there were irregularities with batch entry of data. This is perfectly fine provided the actual patient visitation time is recorded (case-date).  Moreover, they were submitting cases of fever. This can be easily explained with the fact that the recruited data entry personnel were required to submit an average number of records per month and it is possible that they were cheating to meet their quota to receive the full pay check. In the statistical analyses tools this would depict an unusual escalation of fever cases for that day. Similar malicious coding of health records related to Measles and Tetanus disease burdens were found in the analyses carried out by the research analyst – Lujie Chen (Auton Lab).

Lujie also observed biases in preliminary diagnosis of certain disease, which manifest in in similar way, for example flu-like symptoms. The Variance in doctors preference can range from common cold, Cough, Respiratory Tract Infection. This issue of doctor preference may dilute signals and reduce the ability to quickly detect emerging outbreaks.

The effects of low quality data invalidating epidemiological surveillance, is a challenge that the RTBP faces (click to view paper on “automated detection of data entry errors“. These findings were presented at the International Society for Disease Surveillance 2010 conference in Park City (UT), USA by our colleagues from Carnegie Mellon Universities Auton Lab.

Click to view the ISDS 2010 data quality slides

A paper on the “challenges of introducing disease surveillance technology in developing countries“; was also submitted to the ISDS 2010 conference.

We had contributed three chapters in the context of the Real-Time Biosurveillance Program pilot (RTBP) -

• Chapter 9: “The role of Data Aggregation in Public Health and Food Safety Surveillance” – Artur Dubrawski

• Chapter 13: “User Requirements towards a Real-Time Biosurveillance Program” – Nuwan Waidyanatha and Suma Prashant

• Chapter 14: “Using Common Alerting Protocol to Support a Real-Time Biosurveillance Program in India and Sri Lanka” - Gordon A. Gow and Nuwan Waidyanatha.

These chapters written in late 2008 early 2009 carry the conceptual design of the RTBP that is now approaching its completion. I hope that we get the opportunity to publish the wealth of knowledge we’ve gained over the two years, implementing and evaluating the RTBP, in a similar book.

The private sector in developing countries is increasingly interested in extending mobile telephony services to low income and rural markets that were previously considered unprofitable. Determining the right price is a central challenge in this context. Despite known limitations, the Contingent Valuation (CV) method, which elicits information on the Willingness to Pay (WTP), is a useful guide to pricing decisions. The present study draws on data generated using the CV method to examine whether mobile use is sensitive to small declines in the current per-minute price of use for low income households in six countries of Asia: Bangladesh, India, Pakistan, Philippines, Sri Lanka, and Thailand. A Heckman model is used to correct for the sample selection problem arising from the study of mobile phone owners alone. We find that demographic criteria, including income, are not significant in explaining whether usage is responsive to price fall, although they appear important in determining mobile phone ownership. Instead, subscription to multiple service providers has an important association with the price sensitivity of use: Those with multiple SIM cards are likely to increase usage when price falls whereas those who report that they would not switch service providers are unlikely to do so. The study further finds that consumption would increase among those with a more diversified use of mobile services (to participate in competitions and to access government services) and among more ‘limited’ users (those who attach a greater importance to the emergency uses of the phone). Overall the findings suggest that there exists a latent demand for mobile minutes among low income households that can be tapped through a small reduction in price. However, given the relatively low profit margins in these markets and the ability of users to switch between service providers quickly and at low cost, competing on price could threaten the long term survival of firms. Non-price strategies would therefore be important for firm survival and sustainable service delivery.

Click to view workshop report

- Mobiles on Health Calls, The Hindu Business Line, September 13, 2010

- Pilot study in using mobile technology for disease reporting shows promise, Thehindu.com, July 07, 2010

– Pilot study on epidemiological early disease warning system, Chennaionline.com, July 07, 2010

- New tech to keep tab on diseases, timesofindia.com, July 08,10

- Cellphone-based technology to build the arteries for healthcare, Livemint.com, Sep 03, 10

- Building mobile health, Thehindu.com , September 17,10

- New mobile technology can cut biosurveillance cost by 50%, Financialexpress.com, Aug 06, 2010

- Detecting disease-mobile way, expressbuzz.com, July 08,10

- Health on the Go, Voice and Data, September 2010

- Cell Phones to Track Diseases, Deccan Chronicle, July 07, 2010

The Real-Time Biosurveillance Program (RTBP) pilot, during the first week of April, interviewed health workers and health officials in Kurunegala District to study the notification and response policy and procedures. These interviews revealed that in some occasions by the time health officials receive the notification to inspect the patient, with the infectious disease, at the patient’s residence, the patient had already died; health workers literally pull their hair trying to decipher the illegible handwriting on the paper forms; they also mentioned that they have to travel long distance from their villages to the Medical Officer of Health (MOH) office to pickup the paper forms with the patient’s information. These inefficiencies and excessive costs can be drastically reduced with ICT; with a technique as simple as a communicating the information via SMS text messages that costs Rupees 0.50 (fifty cents).

We asked health workers as to how they came to know of any adverse health events (i.e. rise in cases of a disease or an outbreak) either in their own or neighboring jurisdictions. Most of them replied “through word-of-mouth” while some said they came to know through the main stream media (TV, Radio, News Papers). Ideally they should be receiving first hand information from the health system itself.

A summary of the RTBP conducted assessments are documented: field report for India and field report for Sri Lanka.

An ICT system for alerting and situational awareness, is what the RTBP is introducing to the health system in India and Sri Lanka. RTBP adopts a global standard called Common Alerting Protocol (CAP) for issuing standardized alerts and situational awareness messages. Besides structuring complete all-media all-hazards messages, CAP provides grounds for interoperability. The CAP guidelines and preliminary design was adopted from the USA Center for Disease Control Public Health Information Network CAP Guide.

The RTBP introduced technology and procedures are a paradigm shift from the present day surveillance and notification system. The new paradigm introduces situational awareness and alerting, giving the health officials advance notice to decide on the actions that would make a difference, opposed to an unproductive set of formalities that are not aimed towards saving lives but just for pushing statistics up the chain.

In the RTBP, when a health event of interest such as an increased number of common syndromes or diseases are identified through the T-Cube Web Interface, authorized health officials issue a message to the restricted recipients (i.e. health workers) in the effected as well as surrounding areas. Then health workers will be aware of the present danger and can mobilize to respond immediately or chose to be vigilant and observe the situation. For example, a health worker in the effected area would know that he/she needs to investigate the cases in their jurisdiction; while health workers in other neighboring areas may conduct awareness campaigns to educate the public.

The CAP messages designed for health alerts and situational awareness are delivered via SMS, Email, and Web. The single input multiple output software system was developed by Respere Lanka and is embedded in to the Sahana Messaging and Alerting Module. Sahana is a Free and Open Source Software specifically designed for managing disasters.

These health workers (few of them are in the photo with their backs to you) were given training and mobilized with the mHealthSurvey, mobile phone application, for submitting patient disease/syndrome data for the surveillance of epidemiological events. Data that used to take over 15 days to relay up to the paper chain, but was not subject to any detection analysis (i.e. just reporting), now takes several seconds. Moreover, the RTBP collects all communicable and non-communicable diseases along witht their syndrome opposed to a handful of diseases (i.e. Integrated Disease Surveillance Program S and P list of diseases). Each Primary Health Center sends over 100 patient records (probable, suspected, and confirmed cases) a day that is now subject to, RTBP introduced, real-time health event detection analysis. Although there were some errors due to misspelling at the begining, once they were asked to be cautious and were made aware of the consequences of the errors resulting in false statistics that may lead to false alarms of disease outbreaks, they have reduced the error rates to almost zero.

Dr. Ganesan M., (Senior Program Officer, RTBI – extreme left seated at head table, facing you, talking the Health Workers in the photo), present the paper titled: “Real-Time Bio-surveillance Program: Field Experience from Tamil Nadu, India” at the 7th Indian Association for Social Sciences and Health (IASSH) conference on Health, Poverty and Human Development held at Banaras Hindu University, Varanasi from 5th to 7th March 2010. Dr. Ganesan is part of the research team at the Indian Institute of Technology – Madras’s Rural Technology and Business Incubator (RTBI) conducting the RTBP action research in the state of Tamil Nadu, India.

View the conference presentation slides

Read a brief on the conference participation