Smart malaria microscopic diagnosis

In Uganda

Excelscope is a low-cost device that uses the abilities of a smartphone to operate as a smart microscope for Malaria Diagnosis. It is designed to be partially locally produced and repaired with no electronic components. It consists of a smartphone lens module that when reversed provides magnification. The images captured by the device are processed by AI which enables healthcare staff to provide a reliable and timely diagnosis.
The design was developed by a team of two product and two interaction designers in a period of 6 months. It included two field trips, the first one for gaining understanding of the current healthcare system in Uganda and the second for testing the design concept. the design has been awarded, nominated and exhibited in multiple events.
> Medical Delta Young Talent Award 2017
> Delft Global Initiative seed fund

> INDEX design to improve life. Category: Body 
> Lapiz de Acero, Category: Concept-Product

> International Festival of Technology Delft - IFoT 2017
> Mind the step, Dutch Design Week 2017
The problem
In low resources settings, as Uganda, and especially in healthcare field, there must various design opportunities. However, it is not clear where a design intervention could actually make a difference in such a complex context, therefore, one of the biggest challenges of this project was to clear the design fuzzy front and identity through design thinking a clear problem that framed who, where, what and why.
With a qualitative analysis of the first field trip, a mapping of current Uganda healthcare system issues was designed. This system map detailed not only the problems in each level of the healthcare system but also the actors and relations between them.
This map along with literature research showed that besides the poor access to healthcare in the rural area, the health centers lack trained staff and equipment’s, resulting in misdiagnosis. This cause nurses and doctors to prescribe wrong medication, contributing in long term to medicine resistance, logistical problem of medicine shortage and price inflation. Hence, there is a need to equip these health centers to provide a more reliable diagnosis.
Design goal
Choosing the health care 2nd level, the problem was framed in a specific context and user. Additionally, Malaria was selected as a case of study since malaria stands out as 47.3% of the diseases diagnosed in the primary health care and according to the WHO it is the third cause of death in Uganda for both children under five and general population. Thus the design goal was defined as: 
To design a product service system to increase the capability of the nurses in Health Center 2 to diagnose commonly occurring diseases – Malaria- in a reliable, timely and affordable way.
Approach and process overview
Designing for a low resource setting is rather complex given the culture differences, limited resources, level of education and policies among others. Therefore, a human centered approach, accompanied of a deep understanding of those context variables is especially relevant. That is why this project is rooted in direct field research for both problem definition and user testing, which are complemented with design thinking methods.
Besides system thinking and qualitative research, multiple methods were used during the process such as: scenarios to envision potential interactions between the healthcenters and the cloud, personas to promote empathy with the users, literature research into most common diseases and malaria, task analysis to deep into malaria microscopic diagnosis, functional analysis to detail microscope operations and features, benchmarking of similar solutions in the market, and a data journey along with a journey mapping. among others. The latte rcomplemented the human perspective with the process of data transference, this way making possible to frame the diagnosis steps to work on and to clarify which data was relevant and missing from the current process.
Final design
After multiple iterations and a full test in the field, with very positive response, the design was iterated. It is a Product Service System that equipped with the Exelscope is expected to provide the accuracy of microscopic diagnosis by low cost, without a deep previous training and in an automatic way.
The device is equipped with a reverse lens arrangement that provides a high-resolution image with a very low cost, considering that in microscopes the most expensive part is the lense. Users are guided through the app to insert the slide and take a picture of the sample, which is afterwards analyzed by an Artificial intelligence system, currently developed by Makerere University in Kampala, this AI is capable of automatically detects parasites in the sample.

The application guide users thorough a simple linear menu from scanning to the result. It also collects data on patient information, based on current used health information system, HIS II protocol, this in order to match with the  system. The data collected could also be very helpful for research in both malaria and demographics.
Currently the project is still being developed by multiple faculties at TUDelft along with Leiden University.

Methods and tools 
If you feel like checking in more detail which design methods and tools were used for the project, here you go! Because design is not only about the result but about the process 
Field user research
The research was planned to understand the problems faced by the clinical staff in different levels of healthcare centers in primary Healthcare focusing on diagnosis phase. Two major types of experts were interviewed using semi structured interviews: On-field experts; who practiced medicine as their profession in healthcare centers and Off-field experts; who were addressing healthcare issues from as outsiders.

The collected qualitative data was analyzed by the team, this by clustering quotes and insights with three criteria: the level of care facilities, the source of the information, and the relationship with our project. This exercise  gave as a result the problem mapping presented above as well as multiple insights in how the system works and a persona.
Why Healthcare center 2 ?
Choosing which specific spot of all the system problems to work on, the team organized an intern debate to define pros and cons of an intervention in each level of the healthcare system, also factors as such as our skill-set, time at hand, potential clients and opportunities were considered. This way HC2 was chosen due to its impact in reducing above level health centers burden.
After deciding to focus in the HC2 as the problem context, a persona of the nurse working in this place was designed, this to define some factors as level of education, routines and trigger empathy in the design team during the process.
Data and customer journey
Based on a diagnostic process overview, a more detailed version of a malaria diagnostic process was researched and framed. Here, as the project is technology driven, emphasis was given to how the data is being gathered, translated and interpreted. This data journey was developed along with the customer journey to present both perspectives at the same time.
Data journey was complemented with all the ‘frictions’ (internal issues) and ‘noise’ (external issues) that might arise in the process. This provided a very clear demarcation of problems that allow the team to select a specific part of the journey, it means the testing part, in which it is important to provide a certainty and timely results, that are not considered yet.
Multiple scenarios were explored so as to find missing opportunities of the envisioned interactions between both the Health Center (HC) and the diagnostic testing (LAB). The scenarios help to foresee opportunities and drawbacks in the interactions between HC and LAB. The exercise considered two main possibilities to overcome accessibility limitation between the health center and diagnostic testing (LAB). The first case contemplates that the HC components, patient and caregiver are able to move towards the LAB. In the second case the diagnostic testing entity (LAB) and its components, physical and knowledge, are able to move towards the user.
Literature research for disease selection
During the heuristic understanding of the healthcare system, it was found that low level Health Centers main goal was to diagnose and treat the most common diseases for the purpose of providing timely attention and reducing the overwork in high level centers, however these centers are only equipped with nurses, not doctors, so to support the diagnosis process they have been supplied with strips for the easy diagnosis of those common diseases. 
A quantitative analysis was then performed by assigning points to each disease for causes of death, burden of diseases and most commonly diagnosed diseases in primary health care, finding malaria as the most relevant.
Aditionally, Malaria Rapid Diagnostic test (RDT strips) and blood smear microscopy stood out again as 64.2% of the performed test among all the diagnostic tests performed. This lead us to select malaria as case of study disease, and considering that microscopic examination is the gold standard for its diagnosis due to its high accuracy, microscope was our starting point. In this matter there are issues in malaria diagnosis that should be addressed as its misdiagnosis and over-diagnosis
Task analysis
In order to have a better understanding on how is malaria diagnosed by using a microscope, a hierarchical decomposition task analysis was done, it considered the consumables involved, the level of expertise, the steps in which the process might lead towards errors, and gain insights for the specific step in which the microscope is used as a diagnosis tool.  Given the restricted access to the real context, prescribed tasks were researched from the World Health Organization guideline for Malaria Microscopy.
Some crucial insights for the concept development resulted from the analysis. As the difference when preparing a thick and thin slide, the later requires more expertise and the quality of it is imperative for a trustable result and the multiple and complex staining processes, which showed the importance of providing support and instructions to ensure a quality sample.
Functional analysis
To understand the function of a microscope and to identify the essential and nonessential components a functional analysis was conducted. Each part and component of the microscope was studied. The analysis provided key insights and helped in identifying the elements that were most essential for the function of microscope. We also discovered that most components in a microscope are meant for providing freedom of movement, which is required when viewing different specimens with different magnifications. Besides lenses are the most expensive part of a microscope.
Creative session
Once sufficient insights were gathered from the synthesis of the field research and the analysis of related information that followed, a semi-structured ideation phase followed. Two distinct ideation sessions were organized based on creative facilitation models and approach.
The first ideation session comprised of the project team members itself. An approach of describing suitable How-to’s for the larger design goal was defined. Rapid ideas were generated, with the initial obvious ideas being purged, paving the way for more creative and innovative ideas. Once sufficient ideas were  generated, they were clustered suitably to identify the most useful ideas for us.
A similar session was also conducted with a group people from other disciplines to ensure that ideas outside of the team’s bubble and perception were also explored.
Detailing and prototyping
The idea of the first design was to replace the expertise required for the microscope examination with a remote diagnosis support. It consisted of 3 main parts, the preparation module to support the sample preparation, the excelscope as microscope and remote diagnosis tool and the app embedded in it.
Given the complexity of the system where the service is located we identify the possible failure steps by using Failure Mode Effect Analysis (FMEA). This helped in formulating the risk of various touchpoints, severity of failure, their frequency of occurrence and possibility to detect the failure through quantified measures.
After taking the sample and staining it the nurse can introduce the sample into the device which would automatically scan the sample and take a picture, this one is then send to the cloud where a pathologist in any other part of the world is able to examinate. The service assumes that with the right indications the nurse is able to take the sample, stain it and interact with the app embedded in the device.
Preparation module 
This concept consisted of two parts: a preparation module and an examination device. The preparation module allows nurses with low training to be able to perform the task of preparing a good quality blood smear.
Taking as a reference projects that used camera phone to microscopic magnification it was found that reversed lenses served for this purpose, what also served the context of low operational expertise and more importantly lower cost. However, it was understood that further magnification would be needed to detect malaria parasites.
Based on the remote diagnosis service map we developed the application. The features included: new test which considered patients information input, sample pictures and results in a linear menu and test records stage. Android UI guidelines were used since it is the most common system in Uganda. Two iterations with paper prototyping were done before the field trip one with handmade wireframes and one with digital screens.
Field user testing
With the prototype of the design concept we traveled to Uganda for two weeks. As the first field trip, it was divided between on-ground users (nurses and lab-technicians) and off-ground experts (technology and medical researchers, industry experts).
The test aimed to evaluate if the product services system was easy to use for the intended user and if it fits into the context.  The field trip data was collected from different sources, observations, semi-structured interviews and of course the prototype testing. The prototype was tested with multiple users, doctors, nurses and lab technicians in different health center. Both qualitative and quantitative information was collected during the test. The quantitative included Attrakdiff scales to measure user’s perception, System Usability scales and one to ten scales to measure the easiness of the steps.

Check the video for a fun overview of the testing! 
Data analysis
The information collected during the field testing was analyzed to be translated to insights, the most relevant were located according to the step of the process they were relevant and considered the behavior and person involved.
The insights outside the process of diagnostic were divided in 4 fields: strategic, product design, user experience and artificial intelligence.
The most important insights from the field were that: there is definitely lack of equipment, but nurses have more knowledge about the staining than expected, Field stain to is used instead of Giemsa because it is easier and cheaper. Because of the high workload, the lab technicians who do the microscope examination only exam the thick smear in order to fasten the process. Lab technicians get tired after a while and do not check all the sample so reducing the accuracy of the method.
Based on all the insights and the findings through user test and interviews, the following aspects are recommended: change of stain for the local used one, replace remote diagnosis with Artificial Intelligent system, and to do only thick smear.
Overall, the design responds to a clear need in the field, in more detail it was essential to improve the accuracy of malaria diagnosis and to reduce the expertise to do so. However, that is not the only issue around malaria diagnosis, the time for doing it resulted to be remarkable important for both the caregivers and patients. 

You may also like

Back to Top