Malaria is a global disease found in tropical and sub-tropical regions of the world. Despite international initiatives aimed at eradicating the disease, malaria is still responsible for more than 600 000 deaths every year – most of them children.
Malaria is transmitted to humans via the Anopheles Mosquito. The mosquito bite introduces the parasite into the human’s blood via sporozoites in the saliva. The parasites then migrate to the liver, where they mature and reproduce. The organisms multiply in the liver in infected hepatocytes. These differentiate into thousands of merozoites, which rupture the host cell, infiltrating the blood and infecting red blood cells.
There are five known types of malaria parasite: Plasmodium falciparum, Plasmodium vivax, Plasmodium ovale, Plasmodium malariae and Plasmodium knowlesi.
Dengue is another mosquito-borne disease but, unlike malaria, it is caused by a virus rather than a parasite. Dengue is spread by several species of female mosquitoes of the Aedes genus, principally Aedes aegypti. It is an RNA virus of the family Flavivirdae. Dengue is common in more than 120 countries, many of them where malaria is also endemic. Around 390 million people are infected per year of whom about 500 thousand require hospitalization and approximately 40 000 people die.
Despite many advances in vector control, diagnostic screening and effective treatments, malaria continues to have a devastating impact on health around the world.
The World Health Organization (WHO) launched a Global Technical Strategy for malaria 2016 – 2030 in 2015 with the aim to reduce global malaria incidence and mortality rates by at least 90% by 2030. Technical solutions include the control of the mosquito vector, improved treatment and prophylaxis regimes, and effective screening and diagnosis. Malaria-endemic regions are usually economically challenged and extremely cost conscious, where accessible, dependable diagnostic methods are urgently needed.
Diagnosis of malaria is typically carried out by manual examination of thick or thin blood smears stained by Romanowski methods at a pH of 7.2. Other tests include rapid testing techniques and PCR. Most patients presenting with malaria symptoms will have a full blood count (FBC) performed. In addition, these symptoms are like those of other infections, including dengue, which can be endemic in the same area.
HORIBA has developed infectious screening flags from full blood count data, based on innovative and contemporary machine-learning techniques and thousands of measurement points. (1) The clinical performances of these flags have been assessed through on-site validation studies performed for malaria (both Plasmodium vivax and Plasmodium falciparum) and dengue with high clinical sensitivity and specificity.
With the aid of Artificial Intelligence (AI), scientists have developed hematology analyzers that can screen out diseases like malaria and dengue in a few minutes, along with normal CBC reports.
HORIBA 6-part hematology analyzer, Yumizen H550 and Yumizen H500 can raise flags of dengue and malaria in suspicious cases without the use of extra reagent.
This instrument has the potential to segregate Plasmodium vivax and Plasmodium falciparum through AI based algorithms and machine learning from instrument generated raw data measurements.
The flagging messages are displayed and can be printed out and transmitted to a laboratory information system (LIS) or middleware.
Flagging scores are displayed to provide the flags’ triggering level and hence help the users to appreciate the flagging rate.
Visual representation of the scoring flags can be seen in the expert data screen on the instrument for additional information:
The infectious screening flags of the Yumizen H500 and H550 were developed using a Random Forests machine learning algorithm.
This is a common technique used in artificial intelligence methodologies, but it has a particular application in bioinformatics and the development of biomarkers of disease.
While the Yumizen H500 and Yumizen H550 instruments produce 37 reported hematology parameters, the raw data that produces these results is far more complex. In particular, the differential channel generates the resistive and optical measurements of each individual cell plus the pattern and position of clusters generated on the scattergram.
Machine learning is able to take this data and explore the variables in the context of different disease states, using these to generate predictor software. These can then be further analyzed using a large cohort of clinical cases to develop thresholds for flagging.
The two predictors above provide a score that is a probability of having a given disease. The threshold for the predictors allows the triggering of the alarm to capture the majority of cases while minimizing false positives.
Although the patterns generating the flags are not always obvious on the instrument graphics, in some cases the infected red blood cells (RBCs) may be clearly seen on the WBC differential scattergram and can be observed on the left side of Lymphocytes cells population.
Especially in the case of malaria vivax species, the presence of infected RBCs can visually confirm the positive sample.
Malaria infection (Plasmodium vivax) is known to cause interference in the total WBC counts due to parasitized RBC’s. The WBC count is automatically corrected by removing the infected RBCs from total WBC count in HORIBA Yumizen H500 and Yumizen H500, whereas in most of other hematology analyzers there is interference in WBC total counts.
►Malaria infection
In this patient there was a very clear population visible to the left of the matrix.
The algorithm triggered a strongly positive flag for Plasmodium vivax which was confirmed by both blood film examination and a Malaria Rapid Test.
► Dengue infection
Here a febrile patient is present with mild neutropenia, relative monocytosis, and mild thrombocytopenia. A positive flag for dengue was generated by the instrument and confirmed by an antigenic test.
There have been several studies conducted on testing the sensitivity and specificity of these flags in different parts of the world which demonstrates a possibility of the effective use of automated infectious flags for screening malaria and dengue infection in a clinical setting. (1)
One such pilot study was conducted for verifying the flags sensitivity across four different corners of India demonstrated a sensitivity rate of P. vivax as 90% & P. falciparum as 100% and specificity rate of 95.01% & 99.59% for P. vivax & P. falciparum respectively. (2)
The thresholds can also be adjusted as per the population data and requirement. A high cut-off threshold is directly proportional to the specificity. These alarms can be used as markers for the type of infection. This information provided by the instrument within minutes can save the cost for testing all the samples and can be used for early screening of these infections.
Note: Clinical co-relation and further testing is also required to confirm the diagnosis.
The novel technology of Yumizen H500 and Yumizen H550 provides a robust, rapid, automated, and accurate platform for screening of malaria in a clinical setting.
Note: For more details, please do visit https://www.horiba.com/int/medical/academy/technology/malaria-dengue-screening
1. Performance evaluation of machine learning-based infectious screening flags on the HORIBA Medical Yumizen H550 Hematology Analyzer for vivax malaria and dengue fever, Parag Dharap 1, Sebastien Raimbault 2 PMID: 33228680, PMCID: PMC7684750, DOI: 10.1186/s12936-020-03502-3
2. Evalution of Yumizen H550 for screening and classification of type of Malaria (Multicentric in house pilot study for verification of Malaria & Dengue Flags on HORIBA Yumizen H550 Haematology analyzers ,2020, India
3. www.who.int/teams/global-malaria-programme
Mandy Campbell, International Hematology Product Manager, HORIBA
Dr. Prakash Suvasia, Scientific & Medical Officer, HORIBA
Analisador hematológico
Você tem alguma dúvida ou solicitação? Utilize este formulário para entrar em contato com nossos especialistas.