Summary: Diagnosing neurodegenerative diseases is notoriously difficult because symptoms often overlap—a patient might have Alzheimer’s, Lewy body disease, and the effects of a minor stroke all at once. Researchers have developed a breakthrough AI model that can detect five different conditions from a single blood sample.
By analyzing protein patterns across a massive database of 17,000 individuals, the AI identified biological “signatures” for Alzheimer’s, Parkinson’s, ALS, frontotemporal dementia, and stroke. The study suggests that a protein profile is actually a better predictor of cognitive decline than a traditional clinical diagnosis.
Key Facts
- The “Joint Learning” Advantage: Instead of looking for one disease at a time, the AI used “joint learning” to identify a general pattern of brain degeneration across multiple disorders.
- Five-in-One Diagnosis: The model successfully distinguished between Alzheimer’s, Parkinson’s, ALS, frontotemporal dementia, and vascular damage from previous strokes.
- Biological Subtypes: The AI revealed that people with the same clinical diagnosis (e.g., Alzheimer’s) often have different biological protein profiles, suggesting that “one-size-fits-all” treatments may not work.
- World’s Largest Database: The model was trained on the GNPC database, the largest proteomics repository in the world for neurodegenerative diseases.
Source: Lund University
The symptom profiles of different neurodegenerative diseases often overlap, and diagnosing age-related cognitive symptoms is complex. A patient may have multiple overlapping disease processes in the brain at the same time.
Now, researchers at Lund University in Sweden have developed an AI model showing that it is possible to detect several neurodegenerative diseases from a single blood sample.
The study is published in Nature Medicine.
Different neurodegenerative conditions can present with similar symptoms, making it difficult to distinguish between them, for example, Alzheimer’s disease and Lewy body disease, especially in the early stages of cognitive decline.
Now, researchers Jacob Vogel and Lijun An, together with colleagues from the Swedish BioFINDER study and the Global Neurodegenerative Proteomics Consortium (GNPC, an international research consortium that has created the world’s largest proteomics database for neurodegenerative diseases) have developed an AI model capable of detecting multiple diseases at once.
The model is based on protein measurements from more than 17,000 patients and control participants, collected from several datasets within GNPC’s proteomics database, the largest in the world for proteins related to neurodegenerative diseases.
“Our hope is to be able to accurately diagnose several diseases at once with a single blood test in the future,” says Jacob Vogel, who led the study. He is an assistant professor, head of a research group, and part of the strategic research area MultiPark at Lund University, Sweden.
Using advanced statistical learning methods and a process known as “joint learning,” the researchers’ AI model was able to identify a specific set of proteins that form a general pattern for diseases involving brain degeneration.
This learned pattern was then used to diagnose different neurodegenerative diseases. Vogel confirms that their AI model outperforms previous models, while also being able to diagnose five different dementia-related conditions: Alzheimer’s disease, Parkinson’s disease, ALS, frontotemporal dementia, and previous stroke.
The study stands out compared to similar research because the model’s results were validated across multiple independent datasets, according to the researchers.
“We also found that the protein profile predicted cognitive decline better than the clinical diagnosis did, and it seems like individuals with the same clinical diagnosis may have different underlying biological subtypes,” says Lijun An, the study’s first author.
Many individuals diagnosed with Alzheimer’s disease showed a protein pattern more similar to other brain disorders.
“This could mean they have more than one underlying disease, that Alzheimer’s can develop in multiple ways, or that the clinical diagnosis is incorrect. However, I don’t think current protein measurements from blood samples will be sufficient on their own to diagnose multiple diseases, we need to refine the method and combine it with other clinical diagnostic tools,” says Jacob Vogel.
At the same time, he emphasizes that diagnostics is not the only application of their model. Many of the proteins that contributed to the AI model point to areas where follow-up studies could lead to a better understanding of the disease-driving processes behind these neurodegenerative conditions.
The next step is to include more proteomic markers using advanced methods such as mass spectrometry to identify patterns unique to each disease.
“We hope to inch closer toward a blood test that can make reliable diagnosis across disorders without aid from other clinical instruments,” says Jacob Vogel
Facts
GNPC
The Global Neurodegeneration Proteomics Consortium (GNPC) is an international research collaboration and a large-scale database focused on studying proteins linked to neurodegenerative diseases, such as Alzheimer’s disease and frontotemporal dementia. GNPC has created one of the world’s largest databases of proteins associated with neurodegenerative diseases, enabling the systematic analysis of large datasets to accelerate the discovery of biomarkers and advance research on brain disorders.
Proteomics
Proteomics involves studying large datasets on how all proteins are expressed in a collected biological sample. It reveals the unique pattern of protein levels – how much of each protein is present – in, for example, a blood sample, which can provide researchers with important clues about biological functions and how diseases develop.
Key Questions Answered:
A: Clinical diagnoses rely on symptoms like memory loss or tremors, which can be caused by many different things. This AI looks at proteomics—the actual molecular “trash” and signaling proteins the brain leaks into the blood. The study found these biological markers often catch disease processes that doctors miss or misidentify.
A: Yes. The researchers found many patients diagnosed with Alzheimer’s actually had protein patterns belonging to other disorders. This suggests many people suffer from “mixed dementia,” and this AI is one of the first tools capable of unmasking those overlapping layers.
A: While the results are a “world first,” lead researcher Jacob Vogel notes that the method still needs refinement. The next step involves using mass spectrometry to find even more specific markers. It’s currently a powerful research and drug-trial tool, with the goal of becoming a standard clinical test in the near future.
Editorial Notes:
- This article was edited by a Neuroscience News editor.
- Journal paper reviewed in full.
- Additional context added by our staff.
About this AI and dementia research news
Author: Anna Elizabeth Hellgren
Source: Lund University
Contact: Anna Elizabeth Hellgren – Lund University
Image: The image is credited to Neuroscience News
Original Research: Open access.
“A deep joint-learning proteomics model for diagnosis of six conditions associated with dementia” by Lijun An, Alexa Pichet Binette, Ines Hristovska, Gabriele Vilkaite, Yu Xiao, Romina Zendehdel, Zijian Dong, Bart Smets, Rowan Saloner, Shinya Tasaki, Ying Xu, Varsha Krish, Farhad Imam, Shorena Janelidze, Danielle van Westen, The Global Neurodegenerative Proteomics Consortium (GNPC), Erik Stomrud, Christopher D. Whelan, Sebastian Palmqvist, Rik Ossenkoppele, Niklas Mattsson-Carlgren, Oskar Hansson & Jacob W. Vogel. Nature Medicine
DOI:10.1038/s41591-026-04303-y
Abstract
A deep joint-learning proteomics model for diagnosis of six conditions associated with dementia
Co-pathology is a common feature of neurodegenerative diseases that complicates diagnosis, treatment and clinical management. However, sensitive, specific and scalable biomarkers for in vivo pathological diagnosis are not available for most neurodegenerative neuropathologies.
Here we present Proteomics-based Artificial Intelligence for Dementia Diagnosis (ProtAIDe-Dx), a deep joint-learning model on 17,187 patients and controls (age of 70.3 ± 11.5 years, 53.2% female), that uses plasma proteomics to provide simultaneous probabilistic diagnosis across 6 conditions associated with dementia in aging.
ProtAIDe-Dx achieves cross-validated balanced classification accuracy of 70–95% and area under the curve of >78% across all conditions.
The model’s diagnostic probabilities highlighted subgroups of patients with co-pathologies and were associated with pathology-specific biomarkers in an external memory clinic sample, even among individuals without cognitive impairment.
Model interpretation revealed a suite of protein networks marking shared and specific biological processes across diseases and identified novel and previously described proteins discriminating each diagnosis.
ProtAIDe-Dx significantly improved biomarker-based differential diagnosis in a memory clinic sample, pinpointing proteins leading to diagnostic decisions at an individual level.
Together, this work highlights the promise of plasma proteomics to improve patient-level diagnostic workup with a single blood draw.
