Artificial Neural Networks
July 1, 2021
AI in Singapore
July 1, 2021IBM AI Principles
IBM has been a pacesetter in advancing AI-driven technologies for enterprises and has pioneered the future of Machine Learning Systems, a type of AI that enables a system to learn from data instead of through explicit programming. As the system analyzes more information, it is possible to develop better models based on that database. Machine learning (ML) is designed to perform with increasingly greater accuracy.
Deep learning, a subset of ML based on artificial neural networks, uses multiple layers to compute higher-level information usually from large amounts of data. Deep learning systems perform a task multiple times, improving the outcome every time due to the implementation of numerous layers that enable gradual improvement. It is a part of a broader family of ML methods supported by neural networks.
Following decades of research and years of experience collaborating with other institutions, and learning from over 30,000 IBM Watson engagements, IBM developed a process that they call The AI Ladder for Successful AI Deployments:
- Collect: the simplification of data collection and accessibility with a platform allows the management of all data types across all origins and destinations.
- Analyze: building scalable and reliable AI-powered algorithms enables a multidisciplinary and collaborative analysis of all data types.
- Infuse: integrating and optimizing systems across different work frames.
- Modernize: driving multiple AI applications and systems to the cloud.
- Organize: DataOps provides the tools necessary to make data readily and securely available across cloud platforms.
Overview
The potential of artificial intelligence (AI) and analytical systems is evident through IBM’s industry experience, technological solutions, digital services, and the application of intelligent analysis into virtually every idea, concept, and situation. IBM’s AI and their Analytics Services organization assists individuals and organizations in getting their data ready for AI analysis and achieve more efficient data-based conclusions. The access to efficient analytical tools and AI-powered systems centered on privacy and risk prevention gives individuals the confidence to develop their projects and solutions with total security.
IBM Watson’s products and solutions give enterprises the AI tools they need to transform their business systems and workflow while significantly improving automation and efficiency.(1)
IBM IT Aspects
IBM has a diverse portfolio of technology-related products and services.
Since 2016, these services have been categorized into different areas like cloud computing, AI, commerce, data and analytics, Internet of Things, IT infrastructure, Digital workplace, and security.
IBM Financial Aspects
IBM strives to build a long-term and stable business model. IBM has evolved through multiple investments in products and services that have long-term growth and profitability potential.
For the fiscal year 2019, IBM reported earnings of US$9.4 billion, with annual revenue of US$77.1 billion. IBM’s shares traded at over $126 per share.
IBM ranked 34 according to the 2018 Fortune 500 rankings of the largest United States corporations by total revenue.(3)
A Supercomputer to Help Clinicians Make Informed Treatment Decisions
IBM’s supercomputer, Watson for Oncology (WFO), aims to provide clinical oncologists with the most accurate practice guidelines and assist them with informed treatment decisions. Comparing a patient’s medical data to the current treatment guidelines and current research, IBM Watson can mark a precise diagnosis and treatment options.
Watson started with the diagnosis of breast and lung cancers and can now diagnose and treat colon, prostate, bladder, ovarian, cervical, pancreatic, kidney, liver, and uterine cancers, melanoma, and lymphoma.(4)
Concordance Between IBM Watson and Real Clinical Practice for Cervical Cancer Patients
A group of Chinese researchers tried to determine the level of concordance between treatment recommendations proposed by WFO and a multidisciplinary tumor board. They retrieve and retrospectively analyzed the data from 302 breast cancer patients. The proposals were divided into ‘recommended,’ ‘considered,’ and ‘not recommended’ groups.
The results were concordant if the WFO and the oncologists categorized their treatments as ‘recommended’ or ‘for consideration.’ The concordance rate of 200 subjects with adjuvant therapy was 77%. However, the rate was 27,5% in the remaining 102 cases. These are promising findings that can serve as a reference point for the inclusion of AI in oncology.(5)
Concordance of Management Between WFO and a Medical Board for Breast Cancer
WFO is an AI support system treatment option for oncologists. A study in China retrospectively enrolled 300 cases of cervical cancer patients and compared the management recommendation of WFO with real clinical practice.
The results showed concordance (when treatment options were designated “recommended” or “for consideration”) 72.8% of the time. Even with its limitations, this study presents WFO as a future decision-support tool in cancer therapy.(6)
AI-Based Conversational Agent to Support Medication Prescribing
The prevalence of preventable adverse drug reactions in hospitals can affect as many as one of every ten patients. The implementation of AI to assist medical professionals could help prevent many of these mistakes. Scientists from IBM Watson Health try to develop software capable of registering clinical staff queries and giving an accurate answer. The group used an interface called Micromedex integrated to Watson Assistant (WA), a machine learning system. The AI recorded more than 100,000 conversations, including questions about dosing and administration. The results showed WA was able to correctly link user’s queries 80% of the time, showing the potential of these types of systems even at this stage of development.(7)
AI and Lung Cancer Treatment Decision: Agreement with Recommendation of Multidisciplinary Tumor Board
This study aimed to determine the possibility of using Watson for Oncology for clinical treatment in lung cancer patients by comparing WFO treatment indications to those of a multidisciplinary team (MDT).
Lung cancer cases were examined using WFO version 18.4 according to four treatment categories (surgery, radiotherapy, chemoradiotherapy, and palliative care). A total of 340 men and 65 women with different types of lung cancer were enrolled. The results showed that concordance between WFO and MDT was 92.4% in all cases. Agreement was particularly good (100%) in stage IV non-small lung cell carcinoma.(8)
Drug Repositioning for Neurological Disorders Through a Network Medicine Approach
IBM WDD (Watson for Drug Discovery) is an AI platform able to extract data from a large number of documents related to the medical sciences, creating connections between genes, diseases, and drugs.
WDD was used for two independent searches: one for genes associated with psychiatric and neurological disorders (PNDs) and another for drugs associated with PNDs. A total of 1588 genes and 722 drugs were identified as associated with PNDs. These results can be applied for the development of drug-gene-disease interaction databases and drug repositioning follow-up screenings.(9)
References