Non-Invasive Artificial Intelligent Olfactory System for Diagnosis of Parkinson’s Disease

April 06, 2022

Blog

Parkinson’s disease is one of the most common chronic neurodegenerative diseases associated with aging. It affects body movement with primary symptoms such as tremor, rigidity or muscle stiffness, bradykinesia, and postural instability. By the time of diagnosis, Parkinson’s disease (PD) often has reached middle or late stage, leading to more complications.

Existing research shows that seborrheic dermatitis, a very common type of skin rash, is one of the premotor symptoms of PD. While, in adults, it generally occurs in places with larger concentrations of sebum such as the face, chest, or back, it has been linked to a secretion of hormones that leads to an increased production of yeast and enzymes experienced by PD patients.

Additional research suggests the sebum on the skin of PD patients exhibits a unique scent, introducing the possiblity of a non-invasive diagnosis using an odor profile.

Researchers affiliated with Zhejiang University, Zhejiang University School of Medicine, Research Center for Intelligent Sensing at Zhejiang Lab, and Tianjin University of Traditional Chinese Medicine proposed a novel approach that uses fast gas chromatography (GC) alongside a surface acoustic wave (SAW) sensor with embedded machine learning algorithms to form an artificial intelligent olfactory system [1].

In the paper, “Artificial Intelligent Olfactory System for the Diagnosis of Parkinson’s Disease” the team suggests the proposed technique to be fast, portable, low-cost, and easy to operate. It uses several machine learning strategies including a support vector machine, random forest, k nearest neighbor, AdaBoost, and Naive Bayes to construct diagnostic biomarker-based models and odour profile models. 

They use single or multiple parameters to identify the significant volatile organic compound featured in the chromatograms. The system includes three modules:

• Gas injection and preconcentration module
• Chromatographic separate module
• Sensor detection module

Detailed operation of the complete AIO system is explained in the research article.

What Do Results Indicate?

For evaluation purposes, the team collected data from 31 Parkinson’s disease patients and 32 healthy controls (HC) for the artificial intelligent olfactory system, out of which 12 PD and 12 HC were used to evaluate the clinical usage of the models. The results show that the AIO system can be used as a clinical diagnostic approach to diagnose PD patients through the smell of sebum.

The results indicated three significant biomarkers — octanal, hexyl acetate, and perillic aldehyde — between the PD patients and the HC group. Using these three VOC biomarkers, the odor profile collected by the AIO system provided the accurate classification of PD and HC at 70.8% and 79.2%, respectively.

However, there were some limitations to the proposed diagnostic approach:

1. The fast GC method employed by the AIO system to separate mixed VOCs has several limitations. In the GC separation, each peak in the graph represents a pure chemical compound with unique retention time. In a practical scenario, though, two or more compounds can have close retention times, which would cause the peaks to overlap in the fast GC separation.
2. To maintain the high classification accuracy, the data distribution of samples needed to be balanced. The controlled equalization of PD and HC samples did not represent the distribution of PD in a clinical environment. This leads to limited employability of the model.

In conclusion, the proposed diagnostic approach through smell offers new possibility for early diagnosis of the Parkinson’s disease. “Compared with olfactory testing, sleep testing, and other solutions, the combination of the AIO system and ML may produce a new method of gaseous-assisted diagnosis of PD with an improved detection speed and a reduced detection cost,” the team explains in the research article.

This AIO system is not the only non-invasive diagnostic approach for potential PD patients. Another recent effort uses body fluid biomarkers — a group of researchers carried out work to propose a blood-based gene-expression biomarker identification for PD diagnosis using two-layer hybrid feature selection [2]. The research paper is published in the ACS Publications under open-access terms.

References

[1] W. Fu et al., “Artificial Intelligent Olfactory System for the Diagnosis of Parkinson’s Disease,” ACS Omega, vol. 7, no. 5, pp. 4001–4010, Feb. 2022, doi: 10.1021/acsomega.1c05060.

[2] J. Augustine and A. S. Jereesh, “Blood-based gene-expression biomarkers identification for the non-invasive diagnosis of Parkinson’s disease using two-layer hybrid feature selection,” Gene, vol. 823, p. 146366, May 2022, doi: 10.1016/j.gene.2022.146366.