October 2026
M	T	W	T	F	S	S
	1	2	3	4
5	6	7	8	9	10	11
12	13	14	15	16	17	18
19	20	21	22	23	24	25
26	27	28	29	30	31

Vanrose Panashe Nyamangodo: A Multi-Task Deep Learning Model For Malaria And Anemia

AI Generated

Oct 3, 2026, 15:25

Vanrose Panashe Nyamangodo: A Multi-Task Deep Learning Model For Malaria And Anemia

Vanrose Panashe Nyamangodo, Data Science GT at Bindura University of Science Education Official, shared a post on LinkedIn about a recent article she and her colleagues co-authored, published in Computer Science and Information Technologies (CSIT), adding:

”I’m thrilled to share that my paper has been published in Computer Science and Information Technologies (CSIT)!

Malaria caused an estimated 597,000 deaths in 2023, with 95 percent in the WHO African Region. Malaria-induced anemia is especially dangerous for children under five and pregnant women.

Yet the parasite and the anemia are usually diagnosed separately, using different equipment and referral chains.

For patients in resource-constrained areas, that fragmentation can mean follow-ups that never happen. In hematological profiling of malaria-induced anemia using deep learning, we asked, what if one model could assess both at once?

What we built: A cascaded multi-task network that fuses a CNN (blood smear images) with an MLP (WBC, infected RBC, and uninfected RBC counts)

A design that builds in the biology: the anemia prediction is explicitly conditioned on the malaria prediction, since the parasite destroys red blood cells.

A lightweight model, trained on a CPU with 16 GB RAM What we found (independent test set):

96.1 percent recall for malaria detection
93.2 percent accuracy for anemia classification
94.2 percent macro F1-score across both tasks

A note on scope: this was built on 193 patients from the NIH malaria image repository, so it’s a proof of concept.

The next step is the one I care most about: multi-site data collection across Zimbabwe and neighboring endemic countries (target: 1,000+ patients), federated learning to protect patient privacy, and external clinical validation with the Ministry of Health.

This work was carried out during my MTech studies at Harare Institute of Technology.

I’m deeply grateful to my supervisors and co-authors, Wellington Makondo and Simbai Zindove, for their guidance, and to my family and friends for their constant support.”

Title: Hematological profiling of malaria-induced anemia using deep learning

Authors: Vanrose Panashe Nyamangodo, Wellington Makondo, Simbai Zindove

Vanrose Panashe Nyamangodo: A Multi-Task Deep Learning Model For Malaria And Anemia