Field of the Invention:
This invention relates to the development of a hybrid recommender system for the patients suffering
from Parkinson’s Disease. Moreover, this system also offers an automatic monitoring and identification
and recommendations of the patient health status by using smart wearable sensors in collaboration with
machine learning techniques. Such type of system offers a reliable, cost effective and timely treatment
to patients suffering from Parkinson’s disease.
BACKGROUND OF THE INVENTION:
Parkinson’s Disease (PD) is a neurodegenerative disease affecting 10 million of people worldwide. Over
the last three decades, the impact has increased by a factor of four as a result of increasing longevity and
persistent disease (Dorsey, Lancet, 17(11): 939-953, 2018), (Marras, NPJ Parkinson’s Disease, 4 (1): 1-
7, 2018). PD generally occurs at the age of 50 and above in adults, despite the fact that some of the
cases with the young onsets at the age of 18 also diagnosed with PD (Parkinson’s Foundation, USA,
2019).
The biggest challenge faced in the identification of Parkinosn’s disease is the lack of visibility of
symptoms and unavailability of specific test for the diagnosis of PD. The patients totally depend on the
neurologist for its identification and monitoring who takes decisions based on their physical
examinations. Such process is very time consuming and expensive in nature. To resolve such challenges,
integration of smart wearable technologies provides a smart way of monitoring patient’s health
symptoms without visiting to the hospitals. Further integration of smart intelligent technologies offers
fast and effective diagnosis of disease that assists physicians in offering timely treatment to patients.
US2020128801AA describes a method for the identification of Parkinson’s disease through the
measurement of metabolite generated by the Proteus mirabilis strain, and α -synuclein in a biological
sample.
US 20190110754A1 describes a system for identification and monitoring of the neurological disorder.
This system aims to provide a correct assessment of the presence of neurological disorder and its
severity in the patients without requiring any assistance from any trained neurologist.
In order to offer better recommendations to the patients suffering from PD, identification of PD and
knowing its severity level is the major concern. In most of the developing countries like India, where
most of the population living in villages healthcare services are out of reach of people. Inclusion of
smart sensors based technology offers a key solution to most of the healthcare issues and offers better
source for most of the people suffering from chronic disease that are unable to move to the hospitals for
their routine checkup. In the same way, it offers a better opportunity for the people suffering from PD to
regularly monitor their health symptoms without visiting to the hospitals. Thus, research and
development in the integration of sensors based technologies with machine learning technologies are
highly effective in finding an effective solution for providing better healthcare services to larger section
of society.
After surveying lots of papers, in this area, we concluded that the gait related variations are quite
common in most of the PD patients, and are visible at an early stage and increasing with the passage of
time. So, it can be counted as an effective biomarker for early identification of PD. Therefore, the
present invention focus on the analysis of gait cycle captured through smart sensors by using effective
machine learning techniques in order to know the severity level of PD for providing better
recommendations to patients.
Summary of the invention:

Many of the aged people leaving alone suffering from various neurodegenerative diseases faces lots of problems like sudden fall, inability to move to the neurologist for regular checkup and improper care. Such a situation may demand a solution that can deal with such types of issues. One of the key issues addressed by this invention is to suggest a recommendation system to provide an easy, cost effective and efficient way for automatic monitoring, diagnosis and recommendations that will further reduce the regular visits of patients to the neurologists for regular checkup by involving sensors based technology and offering fast diagnosis by integrated the system with machine learning techniques.
The present invention comprising of three stages:
a) First stage is responsible for the classification of subjects into “subjects with PD” and “non PD” classes.
b) The second stage is responsible for knowing the severity level of PD. The result got from first stage passed to the second stage that means only the subjects suffering from PD passed to the second stage that will reduce number of subjects passed to second stage and also reduces computational time and cost.
c) After knowing the severity level of PD, effective recommendations must be given to the subjects depending on their severity level.

In yet another manifestation, the present invention also relays on movement disorder of subjects. As motor and non-motor are the major symptoms that appear in PD patients. The present invention focus on movement disorder as the movement disorder is counted as the early symptom and progresses with the passage of time. Movement disorder is highly observed through gait cycle. Said process comprising of steps:
a) Gait cycle is captured through VGRF (Vertical Ground Reaction Force) sensors that are embedded in the insoles of shoes
b) Captured gait cycle is passed to the proposed system for effective and accurate diagnosis of disease.

We Claims:

1. A system for effective recommendations and identifications of people suffering from PD along with its severity;
2. The system claim1, where the above said system is trained to diagnose the movement disorder based on gait analysis;
3. The system claim 2, where the said raw patient data consists of gait patterns captured through force sensors embedded in the insole of shoes;
4. The system claim 1 and 2 as the system comprises the machine learning algorithms named Random Forest Tree for identification of disease and regressor for severity checking;
5. Along with that the system also claims 1 by providing effective recommendations based on severity level;