Analysis of neural networks-based heart disease prediction system

Zoltan Szucs

Heart disease is one of the major reasons for the increase in death rates. Healthcare is one amongst the most important beneficiaries of huge knowledge & analytics, extracting medical data becoming more and more necessary for prediction and treatment of high death rate due to heart attack. The objective of the study is to analyze different research using different machine learning and deep learning techniques to conclude which is more effective.

Machine learning

The value of machine learning technology is recognized well in Health care industry, which has a large pool of data. It helps medical experts to predict the disease and leads to improvise the treatment.

Artificial Neural Network

An ANN is a multi-layer network containing input, hidden and output neurons. It is trained to teach the problem-solving technique to the network using training data, the training data being validated to stop when it is over fitting in the network and to check the performance

Deep learning

DL are systems that make their own decisions based on intelligence. The advantage of DL is that it allows automatic feature extraction and makes feature learning easier. This approach helps to discover the structure in the data.

Performance analysis of Heart disease prediction using ML and DL

Conclusions

Most of the research works used classification methods such as Association rule mining, Naïve Bayes, Decision tree, ANN, and fuzzy logic for predicting heart diseases. From the results, it is inferred that the performance of the classifier is improved with the Feature subset selection.

Neural network is a training method which works similar to the human brain, and it is an effective technique for predicting the relationship between both the input and the output.

Applying machine learning techniques to medical data helps to predict disease accurately.

Deep learning technique is necessary to process the vast and complex data in the medical field.

From the results, artificial neural network is providing the best performance for heart disease prediction.

 

REFERENCES:

1.     Rajamhoana, S., Devi, C. A., Umamaheswari, K., Kiruba, R., Karunya, K., & Deepika, R. (2018). Analysis of Neural Networks Based Heart Disease Prediction System. 11th International Conference on Human System Interaction (HSI). doi:10.1109/hsi.2018.8431153

Ethereum-and-xgboost

 Prediction of Ehereum Price via xgBoost

Introducere.

Ethereum, un blockchain descentralizat, open-source, cu funcționalitate de contract inteligent, a fost propus în 2013 de programatorul Vitalik Buterin. Dezvoltarea a fost finanțată prin crowdfunding în 2014, iar rețeaua a intrat în funcțiune pe 30 iulie 2015, cu 72 de milioane de monede.

Câteva lucruri interesante despre Ethereum(ETH):

Ether (ETH) este criptomoneda nativă a platformei. Este a doua cea mai mare criptomonedă după capitalizarea bursieră, după Bitcoin. Ethereum este cel mai utilizat blockchain.

Unele dintre cele mai importante corporații din lume s-au alăturat SEE (Alianța Ethereum, este o colaborare a multor start-up-uri din bloc) și au susținut „dezvoltarea ulterioară”. Unele dintre cele mai cunoscute companii sunt Samsung SDS, Toyota Research Institute, Banco Santander, Microsoft, J.P.Morgan, Merck GaA, Intel, Deloitte, DTCC, ING, Accenture, Consensys, Bank of Canada și BNY Mellon.

Informații referitoare la algoritmul folosit - xgBoost

XGBoost înseamnă „Extreme Gradient Boosting”. XGBoost este un framework optimizat de creștere a gradientului distribuit concepută pentru a fi extrem de eficientă, flexibilă și portabilă. Implementează algoritmi de învățare automată în cadrul Gradient Boosting. Oferă o creștere a arborelui paralel pentru a rezolva multe probleme de știință a datelor într-un mod rapid și precis.

Caracteristici XGBoost:

Învățare regularizată: Termenul de regularizare ajută la netezirea greutăților finale învățate pentru a evita supraajustarea. Obiectivul regularizat va tinde să selecteze un model care utilizează funcții simple și predictive.

Creșterea arborelui de gradient: Modelul ansamblului de arbori nu poate fi optimizat folosind metode tradiționale de optimizare în spațiul euclidian. În schimb, modelul este antrenat într-o manieră aditivă.

Contracție și subeșantionare pe coloană: Pe lângă obiectivul regularizat, sunt utilizate două tehnici suplimentare pentru a preveni supraadaptarea. Prima tehnică este contracția introdusă de Friedman. Contracția crește greutățile nou adăugate cu un factor η după fiecare pas de creștere a arborelui. Similar cu o rată de învățare în optimizarea stocastică, contracția reduce influența fiecărui copac și lasă spațiu pentru viitorii copaci pentru a îmbunătăți modelul.

Informații referitoare la setul date

Datele constau în total din 1813 înregistrări (1813 zile) cu 7 coloane.

Marimea: 55.3 kB

2015-prezent

Intrebari frecvente

Este Ethereum o investiție bună?

Pe piețele de criptomonede volatile, cu risc ridicat, este important să vă faceți propria cercetare pe o monedă sau un jeton pentru a determina dacă este potrivit pentru portofoliul dvs. personal de investiții. Dacă ETH este o investiție potrivită pentru dvs., va depinde de toleranța dvs. la risc și de cât de mult intenționați să investiți. Rețineți că performanța trecută nu este o garanție a randamentelor viitoare și nu investiți niciodată bani pe care nu vă puteți permite să-i pierdeți.

Cât de sus poate ajunge Ethereum în 2022? 

Este greu de spus. Nu știm ce se va întâmpla ca urmare a Merge pe termen lung, mai ales după ce piața s-a prăbușit la scurt timp după aceea, așa că ar trebui să așteptăm și să vedem.
De asemenea, țineți cont de faptul că criptomonedele pot fi foarte volatile și că prețurile pot scădea și crește.

Ar trebui să investesc în Ethereum?

Aceasta este o întrebare la care va trebui să răspunzi singur. Înainte de a face acest lucru, totuși, va trebui să efectuați propria cercetare.
Nu investiți niciodată mai mulți bani decât vă puteți permite să pierdeți, deoarece prețurile pot scădea și crește.

Bibliografie

1. Nadeem. “Introduction to XGBoost Algorithm | by Nadeem | Analytics Vidhya | Medium.” Medium, Analytics Vidhya, 5 Mar. 2021, https://medium.com/analytics-vidhya/introduction-to-xgboost-algorithm-d2e7fad76b04.
2. saikiranputta. “Ethereum EDA and XGBOOST Starter Code.  | Kaggle.” Kaggle: Your Machine Learning and Data Science Community, Kaggle, 11 July 2017, https://www.kaggle.com/code/saikiranputta/ethereum-eda-and-xgboost-starter-code/notebook.
3. ysthehurricane. “Bitcoin,Dogecoin,Etc Price Prediction -XGBoost | Kaggle.” Kaggle: Your Machine Learning and Data Science Community, Kaggle, 14 Sept. 2021, https://www.kaggle.com/code/ysthehurricane/bitcoin-dogecoin-etc-price-prediction-xgboost.
4. Dannen, Chris. Introducing Ethereum and solidity. Vol. 1. Berkeley: Apress, 2017.
5. Brownlee, Jason. XGBoost With python: Gradient boosted trees with XGBoost and scikit-learn. Machine Learning Mastery, 2016.

NVIDIA DLSS 3

 

Deep learning super sampling (DLSS) is a family of real-time deep learning image enhancement and upscaling technologies developed by Nvidia that are exclusive to its RTX line of graphics cards, and available in a number of video games.  The goal of these technologies is to allow the majority of the graphics pipeline to run at a lower resolution for increased performance, and then infer a higher resolution image from this that contains the same level of detail as if the image had been rendered at this higher resolution. This allows for higher graphical settings and/or frame rates for a given output resolution. AMD FSR and Intel XeSS are similar, but DLSS has always been the king of the upscalers, as it uses more sophisticated machine learning/AI techniques to piece together more accurate frames.
As of September 2022, the 1st and 2nd generation of DLSS is available on all RTX branded cards from Nvidia in supported titles, while the 3rd generation to generation RTX 4000 series graphics cards.
DLSS 3 augments DLSS 2.0 by making use of an optical-flow frame generation technique. The DLSS frame generation algorithm takes two rendered frames from the rendering pipeline, and generates a new frame that smoothly transitions between them. So for every frame rendered, one additional frame is generated.
DLSS 3.0 makes use of a new generation Optical Flow Accelerator (OFA) included in Ada Lovelace generation RTX GPUs. The new OFA is faster and more accurate than the OFA already available in previous Turing and Ampere RTX GPUs. This results in DLSS 3.0 being exclusive for the RTX 4000 Series.
DLSS also requires and applies its own anti-aliasing method. It operates on similar principles to TAA. Like TAA, it uses information from past frames to produce the current frame. Unlike TAA, DLSS does not sample every pixel in every frame. Instead, it samples different pixels in different frames and uses pixels sampled in past frames to fill in the unsampled pixels in the current frame. DLSS uses machine learning to combine samples in the current frame and past frames, and it can be thought of as an advanced and superior TAA implementation made possible by the available tensor cores.
Nvidia offers deep learning anti-aliasing (DLAA). DLAA provides the same AI-driven anti-aliasing DLSS uses, but without any upscaling or downscaling functionality.

DLSS 3: upscaling and generated frame quality

The author of this article tested DLSS 3, using the RTX 4090, in three games: Cyberpunk 2077, Microsoft Flight Simulator, and F1 22. As expected from the successor to DLSS 2.4, DLSS 3’s upscaling looks impressive: on its Quality and even Balanced modes, upscaled 4K looks just as good as native 3840x2160 across all three games. On its second-fastest Performance mode, there is a noticeable drop in water quality in Microsoft Flight Simulator, but in terms of sharpness it is seemingly very close to native 4k. There’s little to no fudging of fine details, no strange outlines around objects, and the AI-based anti-aliasing remains superior to just about every built-in AA option.

For the most part, generated frames look 99% like a traditionally rendered frame. Here’s a randomly chosen, fully AI-generated frame in F1 22 (upscaled to 4K), followed by a successive rendered frame:

 

 

 3840x2160, Ultra High quality, DLSS 3 Quality (generated frame) 

 

 

  

 3840x2160, Ultra High quality, DLSS 3 Quality (rendered frame)

 

The generated frame looks indistinguishable from a rendered frame. Apart from the visual

artifact that can be seen in the timer on the upper-right corner.

  

 

Left: generated frame. Right: rendered frame

 

And to finish it off, some statistics:

 

 

Data analytics transforming college basketball

In basketball, back in the 1990s, no one considered the three-point shot as the best shot in this sport. The majority of players who controlled the game could not shoot from more than a few meters from the basket.

Even Michael Jordan, the best player in the history of the sport, was a mid-range specialist who attempted no more than two 3-pointers per game throughout his career, while the best players today try a lot of long shots per game, frequently choosing corner shots. What's changed? Analytics.

Adam Petway is the director of strength and conditioning for men's basketball at the University of Louisville. He has an MBA with a focus on sports administration, a PhD in sports science, and has previously worked on the coaching staffs of the NBA's Philadelphia 76ers and Washington Wizards. He just furthered his studies through MIT Professional Education's Applied Data Science Program (ADSP).

When he first started in the profession, data analysis was almost non-existent in training rooms, whereas today there is force platform technology, velocity-based training, GPS tracking during games and training, all to get a more objective analysis to help athletes, so data analysis has increased exponentially.

At MIT, Petway attended live online classes with other people from completely different specialties: lawyers, professors and business executives, him being the only strength and conditioning coach, but he believes that the focus on data gave him and his colleagues a common understanding of each other's work.

He learned two important aspects: learning to write code in Python and using unsupervised learning approaches to put data through artificial intelligence algorithms, considering that since sports teams generate a lot of data, coaches must be able to analyze this data using methods that produce practical insights.

Petway finds it very useful to have access to such high-level data science practitioners, as he is now able to create decision trees, data visualization and run principal component analysis, so he does all these things himself, instead of relying on third-party companies to come and tell him what to do, he takes that data and analyze the results himself, thus saving time and a lot of money.

According to Petway, analytics are advancing the area of strength and conditioning well beyond the time when coaches would only instruct players to complete a particular number of repetitions in the weight room. Wearable technology makes it easier to monitor an athlete's average speed and the amount of practice ground they traverse. 

Petway can assess the force with which basketball players jump and land, using data from a force platform, and even calculate how much force an athlete is producing with each leg and he's also able to measure how fast athletes are lifting weights by using a device known as a linear position transducer. He says the main goal is to develop training regimens that reduce the risk of injury while also increasing the performance of athletes.

Ultimately, Petway notes, coaches are primarily interested in just one data point: wins and losses. But as more sports professionals see that data science can lead to more wins, he says, analytics will continue to gain a foothold in the industry. 

ARTIFICIAL INTELLIGENCE IN CHILD HEALTH CARE - AIBO

Vaccination of children is widely used as a medical procedure for health care. However, procedures and actions such as injections and immobilization cause pain and distress to children. Intense pain and distress in medical procedures are known to cause medical trauma and behavioral problems such as anxiety, depression, fear, and non-compliance with treatment in children and their families, and there is a need for interventions and support aimed at alleviating pain and distress from medical procedures in the pediatric field.

Distraction is one of the ways to alleviate pain in medical treatment.

A study was conducted to evaluate the effects of distraction by a humanoid robot programmed to interact with 57 children receiving the influenza vaccine during the vaccination process. The distraction effect was examined using a control group. The results showed that compared to the control group, the children in the intervention group did not stop crying but smiled more during vaccination. Other studies have reported that humanoid robots intervening during the procedure can distract children, help reduce anxiety, cope with stress, and improve behavior during the procedure. Such AI-based distraction, as a non-pharmacological method, can be widely applied in the field of pediatric medicine.

Study details:

One children's clinic in Mitaka City, Tokyo, which provides general pediatric care, was asked to cooperate in the study. A staff member (psychologist or physician) from the National Center for Child Health and Development (NCCHD) visited the clinic once a week as an observer and asked the caregivers and children for their cooperation while in a waiting room (small enough to accommodate 3 to 4 pairs of caregivers and children) exclusively for those scheduled for vaccination. Those who visited the clinic in odd-numbered months were assigned to the intervention group, and those who visited the clinic in even-numbered months were assigned to the control group. After obtaining the caregiver’s consent, children and their caregivers were asked to spend time with AIBO or a stuffed dog during the waiting time before and after the child's vaccination (about 5 min each) in a waiting room, and children’s behavior was observed by professional observers. The total survey time for each caregiver-child pair was about 15–20 min.

The dimensions of AIBO used in the intervention group were 180 × 293 × 305 mm and it weighed 2.2 kg. It was able to move freely around the waiting room. Its eyes were equipped with OLED displays, and it showed rich facial expressions as well as various voices and behaviors. It also responded to stimuli from touch sensors on its head, neck, and back with facial expressions and voices.

The stuffed dog used in the control group was a gray-haired stuffed animal measuring 130 × 320 × 330 mm in length. It could change the angle of its limbs and neck but it did not move spontaneously and was not equipped with any special functions, such as voice or behavior.

A total of 57 children were included in the study, of which 32 were in the intervention group and 25 were in the control group. The intervention group had a mean age of 4.41 years (3–12 years, median 3 years) and 10 boys (31%), whereas the control group had a mean age of 3.96 years (3–9 years, median 3 years) and 14 boys (56%).

In this study, children's pain behaviors were attenuated after the treatment by the interactive play intervention with the AIBO compared to the play intervention with the dog-shaped stuffed animal. After the intervention with a humanoid robot with interaction capabilities during a child vaccination procedure, it was found that the group in which the humanoid robot intervened was more likely to smile than the group in which it did not intervene, but the degree of crying did not change.

Specifications:

The robot dog model incorporates features of autonomy, object detection/recognition, tactile sensing, obstacle avoidance, wireless LAN, short-term memory, and communication through speech.

The researchers considered using the AIBO’s camera to recognize facial expressions and o infer emotional state.

                                                        

Bibliography:

https://capmh.biomedcentral.com/articles/10.1186/s13034-022-00519-1

https://academic.oup.com/jpepsy/article/41/1/86/2580213?login=false

https://us.aibo.com/

 

 

FotoFinder - detectarea timpurie a cancerului de piele

 FotoFinder Systems GmbH, producător de sisteme medicale imagistice pentru dermatologie, anunţă lansarea bodystudio ATBM master. Sistemul permite pentru prima dată utilizarea Total Body Dermoscopy, o dezvoltare avansată a tehnologiei Automated Total Body Mapping (ATBM®). Esenţa sa constă în interacţiunea inteligentă dintre hardware, tehnologia de imagine şi aplicaţiile software speciale, potrivit unui comunicat al companiei.

(Automatic Total Body Mapping) reprezintă standardul de excelență în dermatoscopia digitală, dovedindu-se un instrument extrem de eficient pentru depistarea cancerului de piele, dar și a altor anomalii ale pielii. Acest sistem complex investighează în detaliu tegumentul și determină evoluția în timp a nevilor pigmentari (alunițelor).

Studiile efectuate la nivel global indică o incidență foarte mare a cancerului de piele. De aceea, devine esențială inspectarea periodică a pielii! FotoFinder încorporează cea mai avansată tehnologie care permite analizarea tuturor modificărilor suferite de piele și luarea măsurilor necesare pentru prevenirea îmbolnăvirii.

FotoFinder este un sistem complex, compus din aparatură performantă. Camera foto profesională fotografiază zona afectată și stochează pozele automat în memoria calculatorului pentru analizarea lor prealabilă de către medicul dermatolog. Medicam-ul integrat este dotat cu lentile speciale care măresc leziunile tegumentare de până la 70 de ori, analizându-le în profunzime pentru a determina dacă prezintă vreun risc de îmbolnăvire sau nu.

Aplicaţia software profesionistă, bazată pe inteligența artificială, Moleanalyzer pro, asistă medicii în analizarea şi evaluarea de risc a semnelor din naştere. Sistemul se bazează pe unul din cei mai puternici algoritmi de deep learning care au fost evaluaţi până acum în teste clinice. În cadrul studiului “Man against Machine”* realizat la University Hospital din Heidelberg, algoritmul FotoFinder a înregistrat scoruri impresionant de ridicate pentru sensibilitate şi specificitate, şi poate concura chiar și cu experţi în dermatoscopie în termeni de calitate a diagnosticului.

Toate sistemele FotoFinder sunt fabricate în propria unitate de producţie FotoFinder din Bad Birnbach, Bavaria (Germania). FotoFinder este certificat DIN EN ISO 13485:2016 şi a primit deja mai multe premii pentru corporate management şi design.

FotoFinder integrează programe focusate de investigare corporală. Acestea sunt:

• Bodyscan: Scopul acestui program este de a detecta noile leziuni tegumentare apărute pe corp. Compară automat pozele inițiale cu cele realizate la următoarea sesiune și semnalează modificările intervenite, recomandând spre investigare amănunțită nevii pigmentari cu structură diferită. .
• Dynamole: Acest program contribuie la detectarea prematură a melanoamelor maligne. Analizează în profunzime structura, mărimea și culoarea leziunilor pigmentare și compară imaginile inițiale cu cele realizate la următorul control, contribuind la emiterea diagnosticului final.

Cine ar trebui să își facă o examinare cu Fotofinder?

Dacă vă regăsiți în cazurile de mai jos, programați-vă acum o vizită la medicul dermatolog:

 Aveți pe corp alunițe în număr mare (mai mult de 50)

 Aveți în familie cazuri de cancer de piele

 Vă confruntați deja cu un melanom

 Aveți alunițe de dimensiuni mari

 Ați observat schimbări în aspectul nevilor pigmentari

 Ați observat apariția unor noi nevi pigmentari

 V-ați confruntat cu arsuri solare în copilărie sau adolescență

 Aveți o piele foarte deschisă la culoare (fototip I)

Regula ABCDE pentru identificarea structurilor pigmentare care prezintă un risc pentru sănătate:

A- Asimetria nevilor pigmentari

B- Bordurile sunt neregulate

C- Culoarea diferită față de cea inițială

D- Diametru mai mare de 50 de mm

E- Excrescența- umflătură care apare peste structura pigmentară deja existentă

Bibliografie:

https://elos.ro/tratamente/dermatoscopie-digitala/fotofinder/

https://www.google.com/amp/s/financialintelligence.ro/fotofinder-prezinta-generatia-urmatoare-de-sisteme-pentru-detectarea-timpurie-a-cancerului-de-piele/amp/

Anipuppy - nose print identification using deep learning

 

A South Korean company has developed a biometric recognition tool allowing dogs to be identified by their nose prints.

Once pet owners register the nose pattern and general information of their dog into an app called "Anipuppy", the information can be easily recalled by scanning the dog's nose print.

“It's a 3D biometric algorithm based on AI (artificial intelligence) and deep learning that we have now put into smartphones so that you can take pictures of the nose patterns and use it to identify each animal," said Sujin Choi, director of iSciLab Corporation.

With the new technology, which the company says is 99.9 per cent accurate, people who find lost dogs can quickly and directly communicate with their owners.

Making dog registrations easier

Currently, it is mandatory to register pets with a microchip or an external ID in South Korea. However, the country hasn’t seen an increase in registration since the introduction of the pet registration system in 2007.

Only 38 per cent of the nation's 6 million pet dogs were registered, according to a 2020 report by the South Korean Ministry of Agriculture, Food and Rural Affairs. Animal rights experts say pet owners are hesitant as they are concerned about the cumbersome process and potential health problems of microchip implants.

That's where the nose ID solution comes in handy - it's not intrusive and much quicker to administer than inserting a chip. iSciLab has been collaborating with the South Korean government since 2019 to develop and test its nose ID technology for commercialisation. 

The project aims to be completed by 2024 so it can become an official dog identification and registration method for the country's database. The company expects to charge around €14 per dog.

It says that in the future, the technology could be used to identify other animals such as cats, cows and deer.

Bibliography: 

• Korean startup launching a nose-print identification for dogs, an alternate to the microchip embedding process (koreatechdesk.com)
• ‘Each dog has a unique nose’: South Korea tests out 'nose print' ID for national pet registration | Euronews
• Korean AI startup offers free nose print option for dog identification - 매일경제 영문뉴스 펄스(Pulse) (pulsenews.co.kr)