NELSON PROVAZIThe same strategy used to create the filters that screen out spam, the unsolicited electronic messages that invade our e-mail in-boxes, is being used by the team of researcher David Heckerman, senior director of the Research Group on e-Science at Microsoft Research, to develop a vaccine against HIV, the Aids virus. “We realized that to be successful, a vaccine would have to attack specific points of the virus, just like anti-spam screens do to select e-mails”, says Heckerman, during a conference at the Latin American Faculty Summit 2010. The event, held jointly by Microsoft Research and FAPESP from May 12 to 14 in the sea-side town of Guarujá in São Paulo state, centered on the theme “Computer Science : making a difference”. More than 200 Computer Science experts from 13 countries met at this sixth Faculty Summit, to hear presentations of innovative projects in several fields of knowledge. A physician by training who did his doctorate in computer science, Heckerman was one of the people responsible for creating the first program for identifying and screening spam in 1997. “Just like the spammers change their e-mails to get through our screening, HIV also undergoes mutations to deceive the immune system and so manage to reproduce freely”, he compared. The chief difficulty in developing an anti-Aids vaccine is that the virus changes constantly. “However, we believe that there are some regions in the HIV genome that are presumably vulnerable to the mutation”, stated the researcher.
Finding these areas is quite a complex task, because one must map all the possible mutations of the virus and the configurations of the protein HLA (human leukocyte antigens), which is the immune system’s tool for hindering HIV reproduction. HLA invades the virus and removes the epitope, a protein fragment responsible for the HIV virus’ genetic information. “We’re looking for these regions called vulnerable epitopes”, said Heckerman. “Our objective is to develop a vaccine that teaches the immune system to recognize only the vulnerable points along the sequence of the HIV genetic material”.
To do this, more than one hundred researchers around the world are using a tool called PhyloD, developed by Heckerman’s group, to evaluate how HIV behaves as from the moment it infects a person. Computers cross the data of people’s immune system and the evolution and mutation of the HIV in their bodies, thereby showing which genetic characteristics help to fight the virus. The statistics generated to date have resulted in the creation of an experimental vaccine, which is expected to go into trials within six months. “If everything works out, we may have effective results in two years”.
Until an effective vaccine against Aids is found, patients must take their antiretroviral medication according to a strict timetable. However, this is not always entirely complied with. An experience conducted in Peru with HIV patients, coordinated by the researcher Walter Curioso from the Cayetano Heredia Peruvian University, who is also an assistant professor at the University of Washington in the US, showed that mobile phones can help patients to comply with their prescribed treatment. “Even when the drugs are provided free of charge, 88% of the patients don’t treat themselves for various reasons”, he said.
The main reason alleged for not complying with the treatment is forgetfulness, as the patient must take several types of medication during the course of the day. Living far away from healthcare centers and concern about discrimination if they are identified as having the virus were the other reasons mentioned. As mobile phones have become indispensable for most people, the research group decided to resort to SMS (short message service) messages to get the patients to stick to their treatment. Thus, they found an innovative use for a common device.
“Patients were interested not only in getting a reminder to take their medication, but also in something motivating, like ‘now it is time for your life'”, the researcher explained. The phrase works like a code, in that it protects the patient’s privacy. The qualitative research study was conducted with 20 men and 6 women with HIV, who assessed the SMS messages system positively. The experience led to the Cell Pos Project, which sends messages to the registered participants and which was developed by the Peruvian university in collaboration with the American university and with Microsoft Research support.
Similar programs for aiding healthcare have been employed successfully in developing nations such as Botswana, South Africa and the Philippines. “In the Philippines, there was a 90% increase in compliance with treatment among people with TB who got messages via their mobile phones”, he said.
NELSON PROVAZIPrevention system
In Brazil, a research group coordinated by professor Jacques Wainer, from the Computer Science Institute at the State University of Campinas (Unicamp), is working on developing a system to detect changes in images at the back of the eye that indicate some degree of diabetic retinopathy, a disease that can lead to blindness. “Statistics show that 15% of the population is diabetic in Brazil. Of these people, 40% have retinopathy and for 8% of them this poses a threat to their vision”, said Wainer, who also presented his work at the Faculty Summit. Diabetes affects the passage of blood through the retina, because the local veins and arteries are weakened, causing hemorrhages and scars that interfere with vision.
The project, approved in 2008 under the second call for projects released by the Microsoft Research-FAPESP Institute of IT Research, proposes to make it easier to screen patients that should undergo specialized tests. Since 2007, when the Institute was created, the two institutions have invested more than R$3.5 million in 11 Brazilian projects in the fields of health, education, digital inclusion, agriculture, e-government, biodiversity, bioenergy and global climate changes.
The research of Wainer’s group is resorting to an uncommon technique to process medical images, based on discovering points with significant color and texture changes. “The points where there was a color or texture change are equivalent to a word”, said Wainer.
The idea is that each type of point should represent ‘words’ in these images. Each image has, on average, 300 such points. Based on these visual words, the research follows two different lines. One of them works with each type of retina anomaly, based on 8 thousand previously classified images. “The most frequent images in our data set concern the exudate, a liquid with high protein content that is produced in reaction to the damage of tissues and blood vessels”, said Wainer. With this method, one discovers that visual words indicate better the presence of exudates in the image. “It’s a fairly precise technique, but it’s slow, because it has to be adapted to each of the various types of anomalies that may appear in diabetic retinopathies”. The system has 90% sensitivity, i.e., 10% of false negatives in the detection of exudates.
The other line of this research project tries to discover the visual words that distinguish which are the normal images and the abnormal ones, without having to look for particular anomalies such as exudates or microaneurisms. “We’re still researching the key points, which will allow us to address the normal vs. abnormal aspects”, he told us. The system should be fully ready in the first half of 2011.
The needs of small farmers are also being considered in one of the projects initiated in 2007 by the Microsoft-FAPESP partnership. This project, called eFarms: uma estrada de mão dupla de pequenas fazendas para o mundo em rede [e-Farms: a two-way street for small farms toward a networked world], is coordinated by professor Claudia Maria Bauzer Medeiros, from Unicamp’s Computer Science Institute, and conducted jointly with the Guaxupé Coffee Growers Cooperative (Cooxupé), which has approximately 11 thousand members in the states of São Paulo and Minas Gerais. “One of the project’s main objectives, from the social point of view, is to create low-cost data communication infrastructure, to link the farms and the co-op, thereby providing the farms with Internet access”, said Claudia. The project involves computer science and agrarian science researchers. The new software programs that are being created will cross and treat the data supplied both by sensors installed in the field, which are to measure variables such as temperature, humidity and luminosity, and by satellites, which are to provide information such as biomass or conditions of the vegetation.
With the computing tools developed, experts should be able to plan production chain activities better, optimize resources and therefore help small farmers perform their tasks. The producers may also take part in the decision-making process, using the data communication network, supplying information for the experts (the co-op, in this case), and giving and getting feedback on the data. “This is the two-way road in the project title, meaning that the farmer not only gets information, but plays an active role in the entire process of generating knowledge to improve his work”, said Claudia. E-Farms already has several modules in operation and others are to be integrated this very year. “The project has enabled the training of researchers in a multidisciplinary task and training several master’s degree and doctoral students”.
Two of the key challenges of the project, which ended on May 31, but which is to continue with Cooxupé financing, are yet to be overcome. The first is reproducing on some of the co-op farms the data communication network infrastructure tested within Unicamp’s controlled environment. “This involves surveying the land, specifying the infrastructure, and determining where to place the antennae and collection points”. The second challenge is to continue to collect and process the sensors’ data, showing the results collected in the network in different ways. “The collected data can already be viewed in real time on the Web, in graph form. Now we want to take the research forward, including the use of maps”, said Claudia. At present, the work involves testing the network in four properties that are difficult to get to. Based on the results of this stage, the co-op will be able to estimate the costs of implantation on a larger scale, potentially reaching 14 thousand properties.
1. Automatic screening of diabetic retinopathies: information technology to fight preventable blindness (nº 08/54443-2); Type Regular Research Awards; Coordinator Jacques Wainer – Unicamp; Investment R$ 290,966.00 (FAPESP)
2. eFarms: a two-way road for small farms toward the networked world (nº 07/54558-1); Type Regular Research Awards; Coordinator Claudia Maria Bauzer Medeiros – Unicamp; Investment R$ 153.313,60 (FAPESP)