This is my end of the year data science project using machine learning to predict stroke. The goal of this project is on how we can use AI to improve healthcare and I also wanted to remember what I have learned. Follow the links to the full video. I'm so guys, this is the end of the year. Did the science project. Yeah. So it's improving, improving healthcare with artificial intelligence, stroke prediction, using machine learning. Yeah. So a lot of people suffer from strokes. What's the main idea of this project here? So what if we could create a system that people can use to predict if the likelihood of them having struck yeah? So this is the. Is the sample that that was used for training of the gender age. Yeah, we have got type attention if someone ever experienced type hypertension experience and heart disease, if their marriage, residence type urban or rule or rule rule. Anyway, I've got the average glucose level in the BMI is the body mass index and smoking status. Yeah, and this is our target. Whether someone is likely is likely to have stroke for one and 0. If you want to have stroke, yeah. So let's do some, let's do some data visualization. So we have. I've gone through scatterplots here. Yeah, so this is average glucose level against the age. So from this scatterplot you've got the samples that are not blue. I'm meaning. People, people don't have stroke and orange for people have good stroke. So the relationship that we can get from here is that as the glucose level increases and age increases are we have got a lot of samples for people have good stroke. So yeah, this is one of the things that we can pick from the data. Another thing is the body mass index again against age. So as the body mass index increases, we don't see a lot of samples, Yeah, but as the age increases, we see a lot of people getting stroke starting from the age of 40 going on once this side. Yeah, so basically that's it. This is just the basic data data visualization. But this is the data that we cleaned. Yeah, it's creating other features from feature engineering, etc etc. Presenting features So he Yeah, What is what is your gender mean? So what is your age? And let's say I'm 10 years old. What is your body mass index, let's say? Glucose level. It's off 100. Let me talk with 15. Umm. Have you ever experienced hypertension? Yes. Have you ever experienced that disease? But let's do it now. No. Are you married? No. Where this the OK where was smoked? I never smoked. Let's see their prediction for this. OK, so the prediction is you're looking at the information you gave me. You cannot have a stroke anytime soon. Predictions score is unrepresented. Yeah, So the model has predicted that you cannot have. His truck now coming to the features that we provided, looking at the edge, it's 10 and the body mass index is 12 and the glucose level is 15. Now look look at this data visualization that we have here. So we said as a glucose level increases. And the age increases. It's possible for some it's someone stands at high risk of getting his job. So now when you look at the age of 10, my age of 10 is somewhere here. Body mass index is somewhere here. Then the average glucose level is somewhere here. So which means so it is basically someone here you see at least makes sense. And yes, this person is not likely to get this joke. Yeah. OK, now what if we say this patient's age is 50? The Body Mass Index 50. The body mass index of from. Let's see it before. The glucose level is. 100. And. This person's experience had hypertension and had to do this and he stays in the urban. So the guy that predicted doesn't have this one is he smokes. Let's see the prediction that we get. Well then, the sky. It's not likely to get this joke. Maybe because it's because of the age. Yeah, because it says at age increases, someone is likely to get struck like saying he's 80. Get their prediction. Well, let's get it, OK? Yeah. OK, what if we increase the glucose level? Because he said, as long as the glucose levels increases, this guy is likely to get a stroke. So that's another 200. Uh-huh. So now you see. The more doses looking at the given information, you may have your stroke. Try visiting the doctor soon and the prediction accuracy is 63. So when we look at our data, our graphs that we have here, this guy has got an edge of 80. Looking at this already is one of the victims. His body mass index is around 4:20 somewhere here. And the glucose level is somewhere to 5th, so we can see this person is like yeah. To get this job. So what if we came in if this guy was never around 40 years? Put mass in there. Yeah, it's 40 and then. And be the glucose liver are used to 50. Yeah, let's see how. Check now what happens accuracy and goes out to 83%.
Artificial intelligence and digital health are not just buzzwords—they're reshaping how we understand and improve public health. On November 14th, from 15:15 to 16:15 at EPH24 in Lisbon, I'll be chairing a pitch session titled "Applying AI and Machine Learning."
I'm genuinely excited to engage in discussions ranging from using ChatGPT for systematic reviews to detecting unrecognized dementia with deep learning methods. We'll explore innovative approaches like machine learning-powered patient records analysis for monitoring injuries in youth and novel digital interventions for preventing type 2 diabetes.
This session isn't just about showcasing technology; it's about understanding how these advancements can make a tangible difference in people's lives. I look forward to learning from the presenters and engaging in meaningful conversations about the future of AI in public health.
If you're attending EPH24, I invite you to join us and share your thoughts on these critical topics.
#EPH24#ArtificialIntelligence#DigitalHealth#MachineLearning#PublicHealth#HealthInnovationDr Stefan Buttigieg MD MSPH MScEUPHA - European Public Health AssociationEUPHAnxtEPH Conference
Did you know that approximately 1% of the global population suffers from epilepsy? Unfortunately, many of them have the pharmacoresistant form of the illness, which is extremely challenging to detect. However, we can leverage AI to aid in this task.
Curious to learn how? Join my session at the Data Science Summit on June 14th in Warsaw during the AI in Healthcare track, where I will present the work of our team at theBlue.ai to address this challenge. We will discuss how we use Computer Vision and Continual Learning to detect Focal Cortical Dysplasias—small changes in the brain that cause significant harm. You can use the code DSSML24SP20 to get 20% discount on the conference tickets.
#EpilepsyAwareness#AIinHealthcare#DataScience#ComputerVision#ContinualLearning#MedicalAI#HealthcareInnovation#TechForGood
🌟 𝐇𝐄𝐋𝐓 𝟐𝟎𝟐𝟒 𝐒𝐲𝐦𝐩𝐨𝐬𝐢𝐮𝐦 𝐈𝐧𝐭𝐫𝐨𝐝𝐮𝐜𝐭𝐢𝐨𝐧: 𝐏𝐚𝐧𝐞𝐥 𝟐 𝐀𝐈 & 𝐑𝐞𝐬𝐞𝐚𝐫𝐜𝐡 𝐔𝐬𝐢𝐧𝐠 𝐇𝐞𝐚𝐥𝐭𝐡 𝐃𝐚𝐭𝐚
AI has been employed across a spectrum of health research domains, including clinical, genetic, behavioral, and public health research in recent years. It is believed that AI will significantly enhance health research by enabling faster and more precise analysis and identification of patterns within expansive and complex datasets. Consequently, AI is expected to hold the potential to effectively address crucial health challenges, thereby advancing the provision of superior healthcare services. Nevertheless, the integration of AI into health research encounters notable limitations, including concerns regarding the quality of available health data, the incapacity of AI systems to display certain human characteristics such as contextual knowledge, and ethical issues stemming from the black-box nature of AI.
This panel brings together experts at the forefront of AI and health research to delve into the transformative impact of AI on the prospective landscape of research using health data, including Erika Ellyne, Allison Gilbert, Chiara BL Formenti-Ujlaki, Thomas Lampert and Dr. Tina Manoharan. The primary objective is to deliberate upon current developments and applications of AI in health research, discuss prevailing challenges, and explore potential ways for optimizing the utilization of AI in the domain of health research.
✨ Interested? Head to our website for all information and registration 👉 https://lnkd.in/dQT9DPRF. Don't hesitate to save your spot early!
📅 Date: April 25, 2024
🗺️ Location: SPARKS, 60 Ravenstein street, 1000 Brussels, Belgium (1 min walking from the Central Station)
#healthlaw#symposium#regulation#healthcare#AI#lawconference#VUB#HALL#healthcareresearch#EHDS#brussels
Problems that AI and data science are addressing in the field of healthcare:
Predictive Disease Outbreaks: AI and data science models can analyze historical health data, environmental factors, and population dynamics to predict disease outbreaks. Early detection helps public health authorities take timely preventive measures.
Personalized Patient Care: By analyzing patient data (including genetics, medical history, and lifestyle), AI can recommend personalized treatment plans. This tailored approach improves patient outcomes and reduces adverse effects. 👨⚕️
Medical Imaging Analysis: AI algorithms enhance medical imaging interpretation. For instance, they can detect anomalies in X-rays, MRIs, and CT scans, aiding radiologists in accurate diagnoses.
Healthcare Resource Optimization: Data science helps optimize resource allocation in healthcare facilities. Predictive analytics can forecast patient admissions, bed occupancy, and staffing needs, ensuring efficient utilization of resources.
#AIinHealthcare#DataScience#PredictiveAnalytics#HealthcareInnovation#MedicalImaging#PersonalizedMedicine#HealthcareTechnology#HealthTech#DataVisualization#MachineLearning#PublicHealth#HealthData#TechForGood#HealthcareOptimization#AI#BigData#HealthIT#HealthcareResearch#InnovationInHealthcare#DigitalHealth
Data suggests that more than 10,000 identified rare diseases affecting more than 30 million Americans and their families, with similar numbers in other parts of the world. Medical challenges don't respect geographical boundaries.
International collaboration in #clinicaltrials is vital in accelerating medical advancements and finding a concrete cure for such rare diseases. By sharing expertise, data, and resources across the country, researchers can expand the reach of clinical trials, enroll more diverse patient populations, and ultimately bring new treatments to patients worldwide faster.
Looking for premium clinical trial support? Check out how Actu-Real can help - https://lnkd.in/dvePtiSE
Follow Actu-Real for more insightful posts on #healthcare, AI, machine learning, health economics, outcomes research, and real-world data.
Can AI and healthcare unite for a healthier future? Discover insights from Purdue University College of Health and Human Sciences Cody Mullen, Ph.D., Program Director of #Purdue’s online Master of Health Administration. Dr. Mullen discusses the transformative role of AI in healthcare. As medical organizations bridge staffing gaps, AI contributes to diagnostics, mental health support and more. Explore how AI is reshaping the healthcare landscape.
➡️ https://purdue.biz/3HamFF3
🔊 Google#AI could soon use a person’s cough to #diagnose disease!
✔️ A team at Google has developed a machine-learning tool called Health Acoustic Representations (#HeAR) that can analyze sounds like coughing and breathing to detect and monitor #health conditions. Trained on 300+ million of short audio clips, #HeAR could potentially #diagnose diseases like #COVID19 and #tuberculosis and assess lung function.
✔️ #HeAR uses self-supervised learning and achieved promising results in detecting #diseases, outperforming existing models.
✔️ HeAR scored 0.645 and 0.710 for COVID-19 detection and 0.739 for tuberculosis, on a scale where 1 represents a model that makes an accurate prediction each time.
✔️ The tool's broad training #data make its results are more generalizable. This development signifies progress in the field of #health#acoustics, offering non-invasive and low-resource methods for disease #diagnosis and monitoring.
➡️ Now, the team is aiming for #clinical data to generate #evidence of #clinical claim and for a real-world patient-centric value that the #SaMD will bring to the society.
➕ Read more: Nature Publication News 21 Mar. 2024 cit. 628, 19-20 (2024) DOI: 10.1038/d41586-024-00869-0
#aiinhealthcare#digitalhealth#voicebiomarkers#medicalinnovation#healthcaretechnology
🙏 Generative AI based photo powered by Microsoft Copilot.
Pharmacist/Medsearch Zambia CEO/Award-winning inovator/Entrepreneur/Tony Elumelu Foundation Alumni
9moVery interesting, hope to see this being launched