![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F3077216%2F9c4a0a306259450249276fadec590879%2FCN1Pv0GUAAEgFmc.jpeg?generation=1678785995667614&alt=media\")
What is Parkinson's
\nParkinson's, first described by Dr. James Parkinson in 1817 as a \"shaking palsy\" is a chronic, progressive neurodegenartive disease characterized by both motor and nonmotor feature. The term parkinsonism is a symptom complex used to describe the motor features of PD and they are tremor, bradykinesia and muscular rigidity.
FoG![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F3077216%2Fa1e823ec544c8b509238c3aa11cb1961%2Ffog2.webp?generation=1678785920071520&alt=media\")
Freezing of gait (FoG) is a unique and disabling clinical phenomenon characterized by brief episodes of inability to step or by extremely short steps that typically occur on initiating gait or turning while walking. During a FoG episode there are changes that happen to the gait metrics. Different types of investigation modalities have pointed to disturbances in frontal cortical regions, the basal ganglia and the midbrain locomotor regions. FoG has been an important clinical problem due to the fact that it has poorly understood pathogenesis mechanism and poor results of empirical treatments.
\nThe typical FoG episode is easily identified. However, the precise definition of the condition has proven to be difficult. The definition according to the 2010 workshop of clinicians and scientist interested in FoG was “brief, episodic absence or marked reduction of forward progression of the feet despite the intention to walk.” The definition can broken down into episodes termed as \"start hesitation\" where patient cannot initiate gait , \"turn and destination hesitation\" in which patients have arrests in forward progression during walking and shuffling forward with steps that are really small.
\nDue to its poorly understood pathophysiology medication is often ineffective and proposed surgical treatments were not fruitful. So any non-medication and non-surgical treatment would be very helpful as alternative to medication. Presentation of external cues (cueing) such as visual, auditory and somatosensory cueing has been shown to be effective at relieving FoG. Incidental cueing, only at the occurrence of FoG (on-demand cueing), was shown to be more effective than continuous cueing in reducing the duration of FoG episodes.
\nWhy is FoG a Machine Learning problem?
\nAs it is already mentioned above, FoG has great clinical importance as it can lead to frequent falls and injuries in patients with Parkinson's disease. However, due to episodic nature, it is difficult to provoke FOG during clinical examination or in research setting. In research setting, detection of FoG is a cumbersome and time-consuming process. However, several studies have tried to evaluate the use of wearable accelerometers to detect FoG automatically. It can also help in assessing FoG in clinical context and for detecting FoG episodes at home.
\nPrediction of FoG episodes is of great value. It is already mentioned that the treatment options for FoG are very limited. Different types of Cueing techniques have shown some promising results. However, for better results, FoG episodes need to be detected or even better predicted and patients can benefits from on-demand cueing techniques.
\nFoG detection research uses data from accelerometer, inertial measurement sensors at ankle, knee and wait and also plantar pressure sensors. Researchers have attempted to use different types of machine-learning algorithms like random forests, support vector machines, CNN and RNN. A recent study proposed a classification algorithm that is based on transformers and CNN and it brought improvement 0.916 to 0.957 in the AUC metric compared with the baseline with corresponding sensitivity specificity and precision of 0.842, 0.939 and 0.617 respectively.
\n![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F3077216%2F50e0dc96898d64defcedd581ea553000%2FFog1.webp?generation=1678785904782846&alt=media\")
References:
\n
What is Parkinson's
\nParkinson's, first described by Dr. James Parkinson in 1817 as a \"shaking palsy\" is a chronic, progressive neurodegenartive disease characterized by both motor and nonmotor feature. The term parkinsonism is a symptom complex used to describe the motor features of PD and they are tremor, bradykinesia and muscular rigidity.
FoG
Freezing of gait (FoG) is a unique and disabling clinical phenomenon characterized by brief episodes of inability to step or by extremely short steps that typically occur on initiating gait or turning while walking. During a FoG episode there are changes that happen to the gait metrics. Different types of investigation modalities have pointed to disturbances in frontal cortical regions, the basal ganglia and the midbrain locomotor regions. FoG has been an important clinical problem due to the fact that it has poorly understood pathogenesis mechanism and poor results of empirical treatments.
\nThe typical FoG episode is easily identified. However, the precise definition of the condition has proven to be difficult. The definition according to the 2010 workshop of clinicians and scientist interested in FoG was “brief, episodic absence or marked reduction of forward progression of the feet despite the intention to walk.” The definition can broken down into episodes termed as \"start hesitation\" where patient cannot initiate gait , \"turn and destination hesitation\" in which patients have arrests in forward progression during walking and shuffling forward with steps that are really small.
\nDue to its poorly understood pathophysiology medication is often ineffective and proposed surgical treatments were not fruitful. So any non-medication and non-surgical treatment would be very helpful as alternative to medication. Presentation of external cues (cueing) such as visual, auditory and somatosensory cueing has been shown to be effective at relieving FoG. Incidental cueing, only at the occurrence of FoG (on-demand cueing), was shown to be more effective than continuous cueing in reducing the duration of FoG episodes.
\nWhy is FoG a Machine Learning problem?
\nAs it is already mentioned above, FoG has great clinical importance as it can lead to frequent falls and injuries in patients with Parkinson's disease. However, due to episodic nature, it is difficult to provoke FOG during clinical examination or in research setting. In research setting, detection of FoG is a cumbersome and time-consuming process. However, several studies have tried to evaluate the use of wearable accelerometers to detect FoG automatically. It can also help in assessing FoG in clinical context and for detecting FoG episodes at home.
\nPrediction of FoG episodes is of great value. It is already mentioned that the treatment options for FoG are very limited. Different types of Cueing techniques have shown some promising results. However, for better results, FoG episodes need to be detected or even better predicted and patients can benefits from on-demand cueing techniques.
\nFoG detection research uses data from accelerometer, inertial measurement sensors at ankle, knee and wait and also plantar pressure sensors. Researchers have attempted to use different types of machine-learning algorithms like random forests, support vector machines, CNN and RNN. A recent study proposed a classification algorithm that is based on transformers and CNN and it brought improvement 0.916 to 0.957 in the AUC metric compared with the baseline with corresponding sensitivity specificity and precision of 0.842, 0.939 and 0.617 respectively.
\n
References:
\nGreat information in here! Thank you for sharing, @drelias15 🙏
", "rawMarkdown": "Great information in here! Thank you for sharing, @drelias15 🙏" }, { "id": 2192221, "postDate": "2023-03-22T14:15:30.530Z", "content": "Good explanation,thank for sharing
", "rawMarkdown": "Good explanation,thank for sharing", "replies": [ { "id": 2192450, "postDate": "2023-03-22T16:57:28.080Z", "content": "You are welcome. We can team up if you are looking for someone to work with.
", "rawMarkdown": "You are welcome. We can team up if you are looking for someone to work with.", "votes": 1, "replies": [ { "id": 2193554, "postDate": "2023-03-23T11:14:26.730Z", "content": "thanks for the invitation. I haven't decided whether to do this comp seriously…
", "rawMarkdown": "thanks for the invitation. I haven't decided whether to do this comp seriously... " } ] } ] }, { "id": 2187187, "postDate": "2023-03-18T13:37:54.203Z", "content": "Thank you for sharing !
", "rawMarkdown": "Thank you for sharing !", "replies": [ { "id": 2187286, "postDate": "2023-03-18T15:41:17.100Z", "content": "You are welcome. Please upvote if you like it.
", "rawMarkdown": "You are welcome. Please upvote if you like it. " } ] }, { "id": 2191342, "postDate": "2023-03-21T22:29:04.700Z", "content": "I am having difficulty spotting what data to start with and how to merge the entire datas together so i can start analysing before building my models
", "rawMarkdown": "I am having difficulty spotting what data to start with and how to merge the entire datas together so i can start analysing before building my models", "isDeleted": true }, { "id": 2208696, "postDate": "2023-04-04T09:05:22.183Z", "content": "Thanks. It's helpful!
", "rawMarkdown": "Thanks. It's helpful!" }, { "id": 2194252, "postDate": "2023-03-23T20:37:27.950Z", "content": "Thanks you for sharing
", "rawMarkdown": "Thanks you for sharing " } ], "comments": [ { "id": 2294046, "author_name": "Vladimir Simões da Luz Junior", "author_url": "", "post_date": "2023-06-09T17:20:18.503000", "content": "Great information in here! Thank you for sharing, @drelias15 🙏
", "votes": 0, "replies": [] }, { "id": 2192221, "author_name": "Hugo", "author_url": "", "post_date": "2023-03-22T14:15:30.530000", "content": "Good explanation,thank for sharing
", "votes": 0, "replies": [ { "id": 2192450, "author_name": "zedreng", "author_url": "", "post_date": "2023-03-22T16:57:28.080000", "content": "You are welcome. We can team up if you are looking for someone to work with.
", "votes": 1, "replies": [ { "id": 2193554, "author_name": "Hugo", "author_url": "", "post_date": "2023-03-23T11:14:26.730000", "content": "thanks for the invitation. I haven't decided whether to do this comp seriously…
", "votes": 0, "replies": [] } ] } ] }, { "id": 2187187, "author_name": "Jino13", "author_url": "", "post_date": "2023-03-18T13:37:54.203000", "content": "Thank you for sharing !
", "votes": 0, "replies": [ { "id": 2187286, "author_name": "zedreng", "author_url": "", "post_date": "2023-03-18T15:41:17.100000", "content": "You are welcome. Please upvote if you like it.
", "votes": 0, "replies": [] } ] }, { "id": 2191342, "author_name": "", "author_url": "", "post_date": "2023-03-21T22:29:04.700000", "content": "I am having difficulty spotting what data to start with and how to merge the entire datas together so i can start analysing before building my models
", "votes": 0, "replies": [] }, { "id": 2208696, "author_name": "Danh Doan", "author_url": "", "post_date": "2023-04-04T09:05:22.183000", "content": "Thanks. It's helpful!
", "votes": 0, "replies": [] }, { "id": 2194252, "author_name": "Bachar Acherif", "author_url": "", "post_date": "2023-03-23T20:37:27.950000", "content": "Thanks you for sharing
", "votes": 0, "replies": [] } ], "raw_markdown_by_id": { "2181059": "\n\n**What is Parkinson's**\nParkinson's, first described by Dr. James Parkinson in 1817 as a \"shaking palsy\" is a chronic, progressive neurodegenartive disease characterized by both motor and nonmotor feature. The term parkinsonism is a symptom complex used to describe the motor features of PD and they are tremor, bradykinesia and muscular rigidity. \n\n**FoG**\n\nFreezing of gait (FoG) is a unique and disabling clinical phenomenon characterized by brief episodes of inability to step or by extremely short steps that typically occur on initiating gait or turning while walking. During a FoG episode there are changes that happen to the gait metrics. Different types of investigation modalities have pointed to disturbances in frontal cortical regions, the basal ganglia and the midbrain locomotor regions. FoG has been an important clinical problem due to the fact that it has poorly understood pathogenesis mechanism and poor results of empirical treatments. \n\nThe typical FoG episode is easily identified. However, the precise definition of the condition has proven to be difficult. The definition according to the 2010 workshop of clinicians and scientist interested in FoG was “brief, episodic absence or marked reduction of forward progression of the feet despite the intention to walk.” The definition can broken down into episodes termed as \"start hesitation\" where patient cannot initiate gait , \"turn and destination hesitation\" in which patients have arrests in forward progression during walking and shuffling forward with steps that are really small. \n\nDue to its poorly understood pathophysiology medication is often ineffective and proposed surgical treatments were not fruitful. So any non-medication and non-surgical treatment would be very helpful as alternative to medication. Presentation of external cues (cueing) such as visual, auditory and somatosensory cueing has been shown to be effective at relieving FoG. Incidental cueing, only at the occurrence of FoG (on-demand cueing), was shown to be more effective than continuous cueing in reducing the duration of FoG episodes. \n\n**Why is FoG a Machine Learning problem?**\n\nAs it is already mentioned above, FoG has great clinical importance as it can lead to frequent falls and injuries in patients with Parkinson's disease. However, due to episodic nature, it is difficult to provoke FOG during clinical examination or in research setting. In research setting, detection of FoG is a cumbersome and time-consuming process. However, several studies have tried to evaluate the use of wearable accelerometers to detect FoG automatically. It can also help in assessing FoG in clinical context and for detecting FoG episodes at home. \n\nPrediction of FoG episodes is of great value. It is already mentioned that the treatment options for FoG are very limited. Different types of Cueing techniques have shown some promising results. However, for better results, FoG episodes need to be detected or even better predicted and patients can benefits from on-demand cueing techniques. \n\nFoG detection research uses data from accelerometer, inertial measurement sensors at ankle, knee and wait and also plantar pressure sensors. Researchers have attempted to use different types of machine-learning algorithms like random forests, support vector machines, CNN and RNN. A recent study proposed a classification algorithm that is based on transformers and CNN and it brought improvement 0.916 to 0.957 in the AUC metric compared with the baseline with corresponding sensitivity specificity and precision of 0.842, 0.939 and 0.617 respectively. \n\n\n\n\n\n\n \n\nReferences:\n1. DeMaagf G. et al. Parkinson's Disease and Its Management. Retrieved from https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4517533/\n2. Nutt J. G. 2020. Freezing of Gait: moving forward on a mysterious clinical phenomenon. Retrieved from https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7293393/\n3. Kwon Y et al. 2022. A practical method for the detection of freezing of gait in patients with Parkinson's disease. Retrieved from https://www.tandfonline.com/doi/full/10.2147/CIA.S69773\n4. Shalin G. 2021. Prediction and detection of freezing of gait in Parkinson’s disease from plantar pressure data using long short-term memory neural-networks. Retrieved from https://jneuroengrehab.biomedcentral.com/articles/10.1186/s12984-021-00958-5\n5. Sigcha L et al. 2022. Improvement of Performance in Freezing of Gait detection in Parkinson’s Disease using Transformer networks and a single waist-worn triaxial accelerometer. Retrieved from https://www.sciencedirect.com/science/article/abs/pii/S0952197622004729", "2294046": "Great information in here! Thank you for sharing, @drelias15 🙏", "2192221": "Good explanation,thank for sharing", "2187187": "Thank you for sharing !", "2191342": "I am having difficulty spotting what data to start with and how to merge the entire datas together so i can start analysing before building my models", "2208696": "Thanks. It's helpful!", "2194252": "Thanks you for sharing " } }