Considering the current estimates and the global social and economic burden of
neurodegenerative diseases, changes in the manner and timing of a diagnosis of these
diseases are urgently needed as well as in the timeliness with which effective
therapeutic interventions are carried out. The complexity of the molecular mechanisms
underlying neuronal degeneration and the heterogeneity of the population of patients
affected by neurodegenerative diseases present enormous challenges to the development of
early diagnostic tools and systems capable of predicting the course of the disease.
Despite intensive research in the field of pharmacology, surgery and rehabilitation,
neurodegenerative diseases remain chronic progressive diseases without a therapy that can
change the course.
MARKERS-NDD is a prospective, observational, longitudinal study, which aims to collect
data from patients affected by neurodegenerative diseases (NDD) followed longitudinally
for routine examinations performed as part of normal clinical practice. Data collected
from clinical evaluations, movement analysis, brain imaging, neuropsychological and
electroencephalographic assessments, blood chemistry tests will be analysed to carry out
statistical investigations and predictive analyses, also using artificial intelligence
systems, which allow the identification of new early markers of diagnosis and prognosis
of neurodegenerative diseases.
Quantitative movement analysis, with the aid of standard motion capture systems (gait
analysis) and with wearable inertial sensors, is a valid tool both for supporting
clinical diagnostics and to assess the response to pharmacological treatment, and to
monitor the progression of NDDs. From this perspective, the kinematic analysis of gait
and graphic gesture can reveal early alterations of motor features to support the
differential diagnosis and stratification of the patient and allow us to follow the
progression of the disease over time.
Although motor symptoms represent key aspects for differential diagnosis between
parkinsonian syndrome and dementia, alterations of cognitive functions, and in
particular, of executive functions can be also present in the early stages of the disease
in patients with synucleinopathies such as PD and Lewy Body Dementia.
In Parkinson's disease (PD), these cognitive disorders can evolve over time towards a
mild cognitive decline (Mild cognitive impairment, PD-MCI) up to dementia (Parkinson
Disease Dementia, PDD).
The identification of neuropsychological markers that can predict the progress of these
diseases and the risk of conversion in patients with PD is a crucial aspect for the
treatment and management of patients in the different phases of the disease.
In the era of artificial intelligence (AI), with the introduction of AI-driven computer
vision, the human movement can be tracked and analysed in real time by the support of a
simple camera of mobile devices, such as tablets and smartphones, potentially eliminating
the need for additional sensors or specialized equipment. Therefore, an approach such as
telemedicine using AI could offer new possibilities and challenges for remote diagnosis,
telemonitoring and telerehabilitation in neurological disorders such as neurodegenerative
diseases. Movement analysis based on AI-driven computer vision could reduce the number of
patient movements, especially in the advanced stages of the disease, often characterized
by severe disability, alleviate the burden of caregivers who are sometimes elderly or
engaged in work activities, and allow consultations to be carried out in remote areas
that are not easily reachable.
The comparison and validation of these systems with the gold standard of movement
analysis represented by gait analysis with motion-captures and the already validated
analysis systems with inertial sensors, constitutes a key piece in the development of
clinical tools that support remote diagnosis and telemonitoring of therapies
pharmacological and rehabilitation treatments.
Similarly, the application of artificial intelligence systems for the kinematic analysis
of the graphic gesture has allowed the development of various pattern recognition systems
for the automatic recognition of handwriting in different application fields.
Dysgraphic features were related to abnormalities of motor and cognitive functions
revealed by clinical and neuropsychological tests, as well as to neurophysiological
correlates of handwriting-related cortical activity such as electroencephalogram (EEG).
Handwriting analysis system which uses AI systems and involves the execution of validated
neuropsychological tests, with the aid of commercial graphics tablets, carried out as
part of normal clinical practice - such as a simple outpatient visit or during of a
structural neuropsychological examination as a component of the now common
multidisciplinary approach that characterizes the management of patients with NDD - could
be particularly useful in PD-MCI to follow the progression of both motor and cognitive
symptoms. As a cost-effective and non-invasive measure of motor and cognitive
performance, graphical gesture analysis systems could be applied repeatedly over time
without serial or meta-learning effects.
A further key aspect in the early diagnosis of NDDs is represented by voice analysis: the
production of the human voice occurs through complex and synergistic movements of systems
and subsystems (vocal cords, larynx, glottis, oral cavity and more), which can be
influenced by the health conditions of the speaker. In particular, among
neurodegenerative diseases, PD and Parkinsonism involve dramatic, objective and
measurable changes in vocal production, which may include (among others) increased noise
levels (due to incomplete closure of the vocal folds) and loss of voice (dysarthrophonia,
dysarthria and hypophonia). Although the evaluation of speech disorder can indeed be
performed via laryngoscope and video-stroboscopic instruments, these are very expensive
tests, requiring a lot of time and qualified personnel. Voice-based artificial
intelligence systems that make use of commercial systems (mobile devices equipped with
microphones with sound robustness) could allow the analysis of the voice with artificial
intelligence algorithms, to diagnose the disease in its early stages.
A long-term analysis of common laboratory blood chemistry parameters and analysis of
different biological samples, such as fecal samples for microbiota analysis, performed as
routine checks by patients with chronic diseases such as NDDs could allow the
identification of associations and early markers useful in the diagnosis and monitoring
of disease progression.
At the same time, the introduction of increasingly powerful and accurate brain imaging
systems, supported by machine learning systems and artificial intelligence techniques, to
obtain an increasingly accurate etiological diagnosis.
Relevant is that each neurodegenerative disease favours a specific brain network which in
turn is associated with a specific loss of tissue in particular brain regions. Therefore,
the possibility of identifying new neuroimaging markers through machine learning systems
to guide the differential diagnosis and accurately evaluate the course of the disease.
In conclusion, the approach adopted by the MARKERS-NDD study is fundamental to increase
the potential for success of an ambitious strategy that aims to develop markers of
progression of neurodegenerative diseases that accelerate the search for
disease-modifying therapy.