Apple

Published by Daniel at October 1, 2021

When tech companies speak of artificial intelligence (AI), they often refer to machine learning (ML) as a component of AI. Many praised gadget functions, such as image recognition, are driven by a subset of machine learning known as “deep learning.” Several examples of AI are used in Apple‘s software and devices, most of them developed in the past couple of years. Including language translation, on-device dictation, and Siri, a personal assistant for iOS. New features around health include forecasting glucose levels, a fall prediction system, and diagnosis for AMI and atrial fibrillation. All these systems benefit from the Core ML features built into the Apple platform, solving user’s problems across disciplines and improving their experience while protecting their personal information.

Machine Learning and AI

ML is employed to assist the iPad’s software to differentiate between a user accidentally pressing their palm against the screen while drawing with the Apple Pencil and an intentional press meant to provide input. It is usual for users’ usage habits to be monitored to optimize device battery life and charging, both enhancing the time users can spend between charges and protecting the battery’s long-term viability.

Then there is Siri, a feature that any iPhone user would recognize as AI. Siri’s speech recognition and attempts to provide useful responses are all driven by machine learning.

ML is also behind the Photos app’s ability to automatically sort pictures into pre-made galleries or find photos of a friend when their name is entered into the app’s search field.

Then there is Siri, a feature that any iPhone user would recognize as AI. Siri’s speech recognition and attempts to provide useful responses are all driven by machine learning.

ML is also behind the Photos app’s ability to automatically sort pictures into pre-made galleries or find photos of a friend when their name is entered into the app’s search field.

For example, an iPhone can take multiple pictures in rapid succession every time the shutter button is tapped. An ML-trained algorithm then analyzes each image and makes a composite of what it deems are the best parts of every image into one result.

Phones have long included image signal processors (ISP) to digitally and in real-time improve photo quality. Still, Apple has sped up the process by integrating the iPhone’s ISP with the company’s latest artificial learning-focused accelerator, the Neural Engine.(1)

Apple Synonym of Innovation

macOS: is the operating system that runs on Apple’s Mac computers. Current features include an aqua user interface where every window element, text, graphic, or widget is drawn on-screen using spatial anti-aliasing technology. The Finder is a file browser that allows quick access to all areas of the computer through the usage of AI. There are 39 different machine languages in macOS to choose from, and the system can be updated using the software Update preference.(2)

Siri: works with any Apple device, makes it easy for people to accomplish everyday tasks quickly using voice, touch, or automation. When people use Siri to ask questions and perform actions, Siri does the language processing and semantic analysis needed to turn their questions into intents for the devices’ app to handle.(3)

iCloud: This service provides seamless access to users to the content they care about, such as photos, videos, documents, apps, and more.(4)

Machine Learning App: It adapts to shifting information and circumstances by basing its actions on the evidence it collects. The user designs it by teaching it how to interpret data and react accordingly instead of specific reactions to a static set of scenarios ⁽⁵⁾

Healthcare IT: patients can learn more about their symptoms or treatment through iPhone and iPad apps, and doctors can access lab results and radiology images. Doctors view high-quality patient imaging studies on iPad tablets thanks to its high-resolution Retina monitor and efficient graphics performance. Secure technology necessitates a security base installed into the hardware, such as Touch ID, a tool that allows healthcare workers to use their fingerprints as passwords. (6)

ResearchKit and CareKit: Medical professionals and experts worldwide are working together to create an interface that can advance health science by allowing them to do large-scale experiments and transform the way they provide care to patients outside of the lab or the doctor’s office. With ResearchKit, researchers can build apps that enroll more patients and conduct studies on a larger scale. CareKit-based apps engage patients in their care and rehabilitation by allowing them to monitor and share everyday progress from their mobile devices. In that sense, providers have a chance to connect with patients outside regular visits.(7)

Apple as an Investment

Apple is the world’s most valuable information technology firm in terms of sales and the most valuable technology company in terms of gross assets. They released financial statements for the third quarter of the fiscal year 2019, which ended on June 29, 2019. The quarterly revenue of $53.8 billion, an increase of 1% from the year-ago quarter. The board of directors of Apple has announced a cash dividend of $0.77 per share of the company’s common stock. On August 15, 2019, the dividend was paid to shareholders on the books August 12 2019. Operating costs are estimated to range between $8.7 and $8.8 billion. (8)

HealthKit, ResearchKit, and CareKit

The three main frameworks Apple provides in this domain are HealthKit, ResearchKit, and CareKit.

HealthKit offers application programming interfaces (APIs) that focus on providing and reading users’ health data. With the users’ permission, shared data in the Apple Health app can be read by third parties, like their physician.

ResearchKit develops applications that aid in the collection of data for research. It primarily focuses on designing visual consent flows, enlisting the user’s participation in a complex mission, and generating surveys.

CareKit is a collection of apps that can help users get better care at home. The framework provides modules to help deliver personalized care plans and track daily progress to generate trends over time.(9)

Smartwatch in the Diagnosis of Acute Myocardial Infarction

Treatment delays are related to high mortality in patients with acute myocardial infarction. ECG recording should be performed in less than 10 minutes from the first medical contact.

A smartwatch could be beneficial for rapid diagnosis in people with symptoms of myocardial infarction. Therefore, a study was developed to determine the applicability of a smartwatch-generate ECG compared with a standard 12-lead ECG in patients with acute coronary syndromes by comparing ST-segment elevations from both sources. The study results suggest a potential use of smartwatches in obtaining an earlier diagnosis of acute coronary syndromes.(10)

Fall Prediction for Elderly Care Using iPhone and Apple Watch

Falling is the second leading cause of death for elderly accidents and injuries based on the statistical datasets from the Ministry of Health and Welfare. The World Health Organization (WHO) reported falls in 28% to 35% of people aged above 65 and 32 to 42% in the age group above 70 years old.

A study proposed a fall prediction system that uses an Apple watch with a built-in three-axis accelerator to measure elderly movement changes. This study aims to notify family members when there is a potential risk for fall, and if already fallen, the Global Positioning System (GPS) would transmit warning messages to the nearby emergency center.(11)

Smartwatch in the Diagnosis of Acute Myocardial Infarction

Treatment delays are related to high mortality in patients with acute myocardial infarction. ECG recording should be performed in less than 10 minutes from the first medical contact.

Fall Prediction for Elderly Care Using iPhone and Apple Watch

Forecasting Insulin-Glucose Dynamics with Machine Learning algorithms

Glucose levels control in patients with diabetes mellitus type 1 (T1D) is a known challenge in medicine, with only a third of patients consistently achieving target glucose levels. The use of machine learning to forecast glucose levels could improve these outcomes. A study directed by Apple Health developed a hybrid system of AI with a statistical approach on real-world T1D data. The researchers used Apple’s HealthKit to collect information from the participants, including continuous glucose monitoring, basal and bolus insulin delivered, estimated carbohydrate ingestion, and estimated active energy burned of two individuals for 150 days. They compared results with pure machine learning models and pure statistical models. The results showed that, on average, the hybrid model consistently produced more accurate forecasts for up to six hours and produced forecasts consistent with the physiological effects of insulin and carbohydrates. (12)

The Use of AI for the Diagnosis the of Atrial Fibrillation

A group of researchers from Ohio selected 50 patients who had undergone cardiac surgery and were on telemetry. An Apple Watch (AW) that performed rhythm assessments three times a day over two days (6 assessments per patient) was assigned to each participant. After a 30 second reading, the result was reported and transmitted (Sinus Rhythm, Atrial fibrillation, Inconclusive). A PDF of the waveform and a telemetry rhythm strip was created and adjudicated by a cardiologist uninformed of the patient’s past telemetry record

The telemetry system identified 34 of 90 Atrial fibrillation (sensitivity of 40%). On the other hand, the AW-generated PDF identified AF in 84 of the 90 cases (sensitivity of 96%). Despite the potential of this technology, the study results indicate that additional validation is necessary before its adoption to the public market.(13)

The Use of AI for the Diagnosis the of Atrial Fibrillation

References