Time series · 28 September 2026
When not to use a forecasting foundation model
Chronos and similar models are pretrained on many series. A short, well-specified local model is still the right tool for some problems.
Administrator, University of Gujrat · PhD, PIEAS
Dr. Madiha Hameed Awan
PhD in Computer Science | AI & Deep Learning Researcher
Medical AI·Computer Vision·Bioinformatics·Proteomics·Data Science
Who I am
PhD-qualified data scientist and AI researcher specializing in deep learning, transformer-based architectures, vision-language models, and natural language processing. Her work focuses on multimodal and generative AI for healthcare, with particular attention to medical imaging, proteomics, cancer detection, and clinical decision-support systems. She designs end-to-end AI pipelines spanning supervised and unsupervised learning, model optimization, interpretability, and deployment, and is an experienced mentor, instructor, and academic trainer.
What I research
Medical AI
Deep learning systems for cancer detection and clinical decision support across dermatology, neuroimaging, haematology, thyroid, lung, liver, and breast imaging.
Computer Vision
Classification, segmentation, and object detection on noisy, low-contrast, and multimodal medical images.
Vision Transformers
Residual-integrated, channel-boosted, and hybrid CNN–transformer architectures, including ARiViT, ReLViT, CBViT, and related models.
Bioinformatics and Proteomics
Protein-sequence analysis, multi-tissue protein expression classification, and computational approaches to leukaemia-related prediction.
Generative AI
GANs and related generative methods for medical image quality, augmentation, and channel-boosted transformer pipelines.
Explainable AI
Interpretability for medical imaging models, including attention maps and XAI methods for breast ultrasound.
Featured contributions
Published
CCNN-SCD
Dermatological diagnosis of skin cancer from lesion images, where visual inspection is labour-intensive and lesion appearance is complex.
PublishedARiViT
Classification of noisy brain medical images, where overlapping intensities and image noise hinder reliable tumour analysis.
PublishedReLViT + YOLO AML
Classification and detection of acute myeloid leukemia on bone marrow images with complex structure and overlapping intensity.
PublishedISIC systematic review
Fragmented evidence on skin cancer classification and segmentation using the ISIC datasets.
Research pipeline
These contributions are listed with their current scholarly status. They are not presented as published papers.
- MAD-YOLO-CBAM-MSFAccepted
- LungCT-CBTSubmitted
- SEL-MTP-DPCUnder review
- Mad-Thy-ReLViTUnder review
- CBViTWith editor
What I have developed
Project work
Brain cancer detection web system
A hospital-facing platform for brain tumour detection and classification, with CNN and transformer-inspired models.
DeployedUniversity LMS
Led design and deployment of the University of Gujrat learning management system, 2013–2014.
Project workIntrusion detection system
Deep learning IDS comparing MLP, CNN, and transformer sequence models. Project work; not a CV-listed journal paper.
Where I have worked
2014 – present
Administrator
University of Gujrat, Pakistan
Nov 2023 – Feb 2024
Data Analyst Trainer (online)
International Certificate Center, Accra, Ghana
Sept 2019 – Oct 2023
Visiting Lecturer
Arid University, Rawalpindi, Pakistan
Selected publications
- Published2026 · Q1
CCNN-SCD: A deep composite architecture for skin cancer detection using feature engineering-aided convolutional neural networks for dermatological diagnosis
Madiha Hameed, Aneela Zameer, Muhammad Yousaf Hamza, Muhammad Asif Zahoor Raja
Applied Intelligence
doi:10.1007/s10489-026-07111-6 - Published2026 · Q1
Explainable AI in breast cancer ultrasound imaging: current developments and challenges
Madiha Hameed, Kok Swee Sim
Frontiers in Digital Health, 8:1846763
doi:10.3389/fdgth.2026.1846763 - Published2025 · Q1
Acute myeloid leukemia classification using ReLViT and detection with YOLO enhanced by adversarial networks on bone marrow images
Madiha Hameed, Muhammad Asif Zahoor Raja, Aneela Zameer, Hafsa Shareef Dar, Ala Saleh Alluhaidan, Romana Aziz
Scientific Reports, 15, 32798
doi:10.1038/s41598-025-17891-5
From the blog
RAG · 22 September 2026
When RAG returns the wrong passage
A language model can quote a retrieved chunk and still miss the claim. Retrieval quality, not prompt tone, decides whether RAG helps.
Image processing · 17 September 2026
Augmentation does not create new patients
Flips, colour jitter, and GANs can regularise a small set. They do not replace an external cohort or a second scanner.
Teaching and supervision
Attia Noor
Comprehensive diagnostic strategies for thyroid cancer addressing class imbalance with advanced artificial intelligence learning techniques
MS · Completed · Co-supervised · 2025
Hamna Zubair
Advanced feature fusion for artificial intelligence driven breast cancer imaging
MS · Completed · Co-supervised · 2025
Khizra Noor
Deep learning empowered framework for improved diagnosis in thyroid carcinoma
MS · Completed · Co-supervised · 2025
Syeda Muntaha Bader
Explainable artificial intelligence for multimodel imaging in breast cancer diagnosis
MS · Completed · Co-supervised · 2025
Explore
Read the bibliography, the research portfolio, the services, the blog, or get in touch.