*The authors thank their colleagues in the NIA Artificial Intelligence and Data Sharing Working Groups for their support on this post.

Biomedical data science is fast evolving, thanks in large part to the growth of artificial intelligence (AI) and machine learning (ML) technologies as powerful additions to the scientific communitys toolbox. The challenge is harnessing the massive data flow including whats produced by NIA-supported research and making it easier for investigators to tap into. Several teams at NIA and across the broader NIH are working on solutions, and were pleased to announce supplemental funding is now available in four key areas to help researchers modernize their data.

NIH data policy aims to make data Findable, Accessible, Interoperable, and Reusable (FAIR). NIH also aims to ensure that our data repositories align with the Transparency, Responsibility, User focus, Sustainability, and Technology (TRUST) principles. The goal is to have high-impact data usable with AI or ML applications to improve our understanding of healthy aging and identify factors and interventions associated with disease resistance and successful treatments.

In Fiscal Year (FY) 2021, NIA partnered with the NIH Office of Data Science Strategy (ODSS) to supplement active NIA research projects in four key focus areas. The NIA community responded robustly to this opportunity, resulting in 23 supplement awards across these four notices of special interest (NOSIs). ODSS funded 19 supplements and NIA supported four, investing a combined total of nearly $6 million.

In the true spirit of open science, this funding will aid the development of teaching materials, workshops, and freely accessible online content so that other investigators can train their students. For example, these awards support scientists who are creating classes and curricula aimed at making data AI/ML-ready and aligned with FAIR and TRUST principles.

In FY 2022, NIA is again partnering with ODSS and has joined four Notices of Special Interest (NOSIs). Three are reissues of last years notices:

The fourth supplement opportunity is new this year:

If youre as excited about the nexus of IT and healthy aging research as we are, we hope youll apply for these NOSIs to potentially accelerate your projects! If you have questions, please email the contacts listed above or leave a comment below.

Read more from the original source:
Funding to make data ready for AI and machine learning - National Institute on Aging

Paraphrasing tools help bloggers and writers in creating new content from the preexisting content. These tools use the advanced technology of artificial intelligence and machine learning to generate paraphrases. In this article, we will be discussing how artificial intelligence and machine learning are trained to work in a paraphrasing tool. Firstly, lets discuss the components of the paraphrasing task

There are two different tasks in paraphrasing. These two tasks are paraphrase identification (PI) and paraphrase generation (PG).

The purpose of the paraphrase identification task is to check if a sentence pair is pointing towards the same meaning. In paraphrase identification, the system yields a figure between 1 and 0. Here, the value 1 shows that the sentence pair have the same meaning while 0 exhibits that the sentence pair is not a paraphrase of each other. Github: https://github.com/nelson-liu/paraphrase-id-tensorflow.git

The paraphrase identification task is a machine learning task. The systems are always trained with a corpus of sentence pairs. The learned knowledge is then used by machine learning to identify if a sentence pair is paraphrased or not. First, the system is trained with a corpus of labeled sentences pairs. Then, use the learned knowledge to identify whether two sentences are paraphrases

In the second task of paraphrase generation, the aim is to generate one or more paraphrases of the input text automatically. So, the aim is to create paraphrases that are fluent and have the same meaning. The paraphrase identification system takes this task as a classification task whereas the paraphrase generation system takes this task as a language generation task. The algorithms of machine learning (ML) and artificial intelligence (AI) bring about the classification of sentences. It means that these algorithms create a model that is useful for input and output mapping. In other words, machine learning or ML uses a number of strategies to make two sentences similar in meaning.

As the focus of this article is paraphrasing or paraphrase generation, we will now look at different techniques of paraphrase generation. We can classify such techniques into two major categories.

In this approach, the paraphrase generation is controlled by a template or syntactic tree. Kumar and Ahuja with their associates proposed an approach in 2020 for paraphrase generation. It uses both syntactic trees and tree encoders by employing LSTM (long short-term memory) neural networks.

Another approach (retriever editor approach) was given in the same year for paraphrase generation in which embedding distance related to the source is used to select a similar source-target pair. After that, the editor has to modify the input sentence with the help of a transformer. The retriever should select the source-target pair with the highest similarity on the basis of embedding distance with the source. The job of the editor is to modify the input accordingly.

In these techniques, language models with fine-tuning such as GPT2 and GPT3 works to generate paraphrases. There is an approach of paraphrase generation using GPT2 where the ability of GPT2 to understand the language is exploited. GPT2 is trained on a large open-domain corpus, therefore, its ability to understand language is exceptional. The aim of this approach is to fine-tune the weight of the pre-trained GPT2 model.

In this section, we will discuss a unified system architecture that is capable of both PI and PG. The major components of such a system are as follows.

The first component of a system is to collect data from a variety of sources. The sources may be Quora duplicate question pairs, MSRP or Microsoft paraphrase research database, PARANMT 50M, etc. The training set is usually very large because these sources contain a lot of datasets with many thousands of sentence pairs. These different types of data are valuable to train the paraphrasing tool models.

The purpose of this stem is to increase data diversity. It is achieved by sampling and filtering the original data. Usually, paraphrase generation models give correct paraphrases with no recurrence. It is due to the huge lexical resource and syntactic diversity that is present in the data used during training. As a result, the paraphrasing tools generate various paraphrases having the same meaning, however, the vocabulary is varied. In addition, it is necessary to perform a number of transformations to the training data to enhance data diversity. As a result of this step, the diversity semantic similarity and fluency are provided to the system.

The system is trained so that it can perform the task of paraphrase generation. For this purpose, the Text-To-Text Transfer Transformer is used to train the system on data. For instance, it is possible to use T5 based pre-trained model for this purpose which is a Text-To-Text Transfer Transformer.

The models such as the T5 model have a self-attention technique that is used in transformers receiving input sequences and generating an output sequence. The output sequence is of the same length as that of the input sequence. So, it is important to compute every element of the output sequence by performing calculations of an average of the input sequence given.

In the end, the whole model is trained for up to 200 epochs on systems having at least 120GB of RAM (random access memory). The algorithm takes quite a lot of time, about three days, to train on the task of paraphrase generation. The system should be efficient as well as lightweight. It is possible to optimize the parameters to improve the performance of the system further.

There are many paraphrasing tools trained with Machine Learning and Artificial Intelligence.

For example: Paraphrasingtool.ai is the perfect example of AI Based paraphrasing tool that uses its own trained model using transformers to rewrite content. This paraphrasing tool is the most accurate, reliable, free and plagiarism free paraphrasing tool available on the web. It can rewrite content in any language automatically and accordingly. This paraphrase tool is very carefully tested to avoid any manual processing and to ensure the quality.

The process of paraphrasing has two tasks paraphrase generation and paraphrase identification. These tasks have huge significance in NLP or natural language processing. There are different approaches available for paraphrase generation. Artificial intelligence and machine learning play their roles in paraphrase generation and identification tasks.

Various models such as T5 model works for sentence generation. The system developed as a result of these algorithms is trained extensively with various datasets and data sources. Consequently, the paraphrasing tools based on artificial intelligence and machine learning have increased diversity and a huge vocabulary.

Link:
How AI and Machine Learning trained to work in Paraphrasing tool - Techiexpert.com - Techiexpert.com

To print this article, all you need is to be registered or login on Mondaq.com.

As insurers consider augmenting the quoting process withalgorithmic predictive models, including those aided by artificialintelligence, machine learning, and/or robotic process automation(Models) for which core inputs are,or could be considered, a consumer report, one question that mayarise is whether the Fair Credit Reporting Act, 15 U.S.C. 1681-1681x (the FCRA)dictates the distribution of an adverse action notice when a Modelis not implemented for the purpose of makingcoverage and rating decisions(determining whether to accept or decline a particular risk or thepremium charged), but instead for the purpose of determiningwhether other actions can be taken with respect to consumers likerouting applicants to certain payment methods or other designationsunrelated to coverage and rating decisions(administrative decisions).

Under the FCRA, an adverse action can meandifferent things in the context of different industries oruses. In the context of insurance, an adverseaction is defined to mean a denial orcancellation of, an increase in any charge for, or a reduction orother adverse or unfavorable change in the terms of coverage oramount of, any insurance, existing or applied for, in connectionwith the underwriting of insurance."1 Under adifferent section of the FCRA, If any person takes anyadverse action with respect to any consumer that is based in wholeor in part on any information contained in a consumer reportthat person must, among other things, provide an adverse actionnotice to the consumer.2

A consumer report is defined tomean any written, oral, or other communication of anyinformation by a consumer reporting agency bearing on aconsumer's credit worthiness, credit standing, credit capacity,character, general reputation, personal characteristics, or mode ofliving which is used or expected to be used or collected in wholeor in part for the purpose of serving as a factor in establishingthe consumer's eligibility for . . . (A) credit or insurance tobe used primarily for personal, family, or household purposes; or .. . (C) any other purpose authorized [as a permissible purpose ofconsumer reports]"3 Thepermissible purposes of consumerreports include, in relevant part, the furnishing of a consumerreport by a consumer reporting agency to a person which ithas reason to believe . . . intends to use the information inconnection with the underwriting of insurance involving theconsumer."4

First, insurers should consider whether an administrativedecision could be considered [1] an increase in any chargefor . . . or other adverse or unfavorable change in the terms ofcoverage . . . applied for, [2] in connection with the underwritingof insurance.

An administrative decision could be considered an increase inthe charge for coverage, because applicants subject to anadministrative decision could be giving more value for the samelevel of coverage in some way. Such additional value could beminimal to the point of appearing nominal, but could theoreticallybe construed as an increase.

An administrative decision could be considered an adverse orunfavorable change in the terms of coverage, because the burden ofhaving to pay premium in a different way or obtain or interact withtheir coverage in a different way could be construed asadverse or unfavorable from the perspective of theapplicant. In many circumstances, particularly thoseaffecting applicants with fewer resources, paying more at one timeor in a different manner could mean the applicant has less funds onhand to contribute to other needs. An administrative decisioncould therefore be considered adverse orunfavorable.

Depending on the nature of the administrative decision, it couldbe construed as being undertaken in connection with theunderwriting of insurance. The only permissible purpose forwhich a consumer report may be provided to an insurer is touse the information in connection with the underwriting ofinsurance. Further, it seems counterintuitive that thelegislative intent of the FCRA would be to permit the provision ofconsumer reports without the attachment of attendant restrictionsand obligations like the FCRA's requirements in respect ofadverse actions.

As stated above, according to the FCRA, if any person takes anyadverse action with respect to any consumer that is based in wholeor in part on any information contained in a consumer reportthe person must, among other things, provide an adverse actionnotice to the consumer.5 Insurers must thereforeconsider whether an administrative decision could be construed asbeing (1) based in whole or in part on (2) any informationcontained in a consumer report.

The phrase based in whole or in part on has beeninterpreted to apply only when there is a but-forcausal relationship. An adverse action is not considered tobe based in whole or in part on the consumer report unlessthe report was a necessary condition of the adverseaction.6

Under certain caselaw, the baseline or benchmark for consideringwhether there has been a disadvantageous increase in rate (and,therefore an adverse action requiring notice to the applicant) hasbeen interpreted to be what the applicant would have had ifthe company had not taken his[/her] credit score intoaccount."7It may be that the onlypurpose of a Model's use of a consumer report is to determinewhether an administrative decision will be engaged. In that case,the baseline could be considered to be the absence ofthe result of the administrative decision. In other words,without use of the Model that integrates the consumer report, theremight not be any possibility of the administrative decisionimpacting the applicant.

An insurer must analyze whether particularized information usedin a Model has been obtained from a consumer reporting agency basedon the insurer's permissible purpose. An insurer shouldalso analyze whether the information is: (i) a writtencommunication of information derived from a consumer reportingagency; (ii) bearing on a consumer's credit worthiness, creditstanding, credit capacity, character, general reputation, personalcharacteristics, or mode of living; (iii) which is used or expectedto be used or collected in whole or in part for the purpose ofserving as a factor in establishing the consumer's eligibilityfor insurance to be used primarily for personal, family, orhousehold purposes.

Finally, an insurer should consider whether the above analysiswould differ or whether additional considerations arise out ofstate insurance scoring laws promulgated based on the NationalCouncil of Insurance Legislators' Model Act Regarding Use ofCredit Information in Personal Insurance (NCOILModel). The NCOIL Model defines whatconstitutes an insurance score (which is similar tothe FCRA's definition of consumer report), what constitutesan adverse action in respect of such insurance scores(which is similar to the FCRA's definition of adverseaction), and when an adverse action notice must be sent in respectof such adverse actions (which trigger language is similar to theFCRA's trigger language). This analysis will depend on thestate-specific implementation of the NCOIL Model (whereapplicable), or on other related state laws and regulationsaddressing this subject matter (for those states that have notadopted some form of the NCOIL Model).

Of course, in analyzing these issues, insurers should consultextensively with insurance and federal regulatory counsel as to thespecific nature of the administrative decisions, how Models arecreated and used, and what the impact of such administrativedecisions and Models are on applicants and consumers.

1 15 U.S.C.A. 1681a(k)(1)(B)(i).

2 15 U.S.C.A. 1681m(a).

3 15 U.S.C.A. 1681a(d)(1)(A) and (C).

4 15 U.S.C.A. 1681b(a)(3)(C).

6 Safeco Ins. Co. of Am. v. Burr, 551 U.S. 47, 63, 127 S.Ct. 2201, 2212, 167 L. Ed. 2d 1045 (2007). This case is alsosometimes referred to asGeico v. Edo.

7 Id.at 2213.

The content of this article is intended to provide a generalguide to the subject matter. Specialist advice should be soughtabout your specific circumstances.

Here is the original post:
When Might The Use Of AI, Machine Learning, Or Robotic Process-Enabled Insurance Models Result In An Adverse Action Under The FCRA? - Insurance -...

Those working in the health sector will tell you of the patient who's sick - but doesn't want to be a bother, so doesn't ask for help, even though they really need it.

Or the family that is desperately worried about the health of their loved one, who is pretending that's everything's ok, when it's not.

Kathryn speaks with Dr Margaret Sandham, who's spear-headed a study into how machine learning could help in the palliative care sector, picking up crucial symptoms that can mark a change in the health of a patient, so appropriate care can be given.

The research, conducted by AUT, analysed the symptoms of 800 patients at an Auckland hospice, using a combination of statistical tools, machine learning, and network visualisation.

Margaret explains how the data could have application for mobile apps and wearable technology - a much less intrusive way of keeping tabs on the health of a patient, than constant phone calls or visits from health workers.

Photo: Supplied, 123RF

Visit link:
New study looks at machine learning and palliative care - RNZ

BOSTON--(BUSINESS WIRE)--Mytide Therapeutics, a company transforming peptide manufacturing with predictive analytics and machine learning, has raised $7 million in Series A financing. The round was led by Alloy Therapeutics, a biotechnology ecosystem company, and was joined by Uncommon Denominator and the Mytide founding team. As part of the financing, Alloy Therapeutics CEO Errik Anderson will join Mytides Board of Directors. This financing will allow Mytide to scale its AI-enabled Gen2 platform to support cost-effective, scalable, and decentralized manufacturing for a wide-variety of peptide and peptide conjugate applications for therapeutic discovery and personal peptide vaccines (PPV).

Mytides Gen2 platform produces both natural and non-natural peptides 30-times faster than traditional manufacturing practices by eliminating bottlenecks throughout the entire process of synthesis, analysis, purification, and lyophilization. Through rigorous in-process data collection, Mytides continuously learning AI-guided engine enables higher purity, production reliability, and speed, by controlling a proprietary set of chemical processes, analytical tools, and robotics. These tools enable access to a novel peptide space including difficult-to-manufacture non-canonical amino acids, constrained peptides, and short-proteins that are inaccessible or uneconomical to produce and screen using traditional peptide manufacturing processes.

Mytides robust data capture and processing techniques represents one of the largest and fastest growing peptide manufacturing data repositories in the world. Through unparalleled manufacturing speed and precision, Mytides technology has addressed the high-throughput screening and library generation needs of computational biology modeling to support in vivo and in vitro studies, as well as clinical trial studies.

At Mytide, we aim to overcome the time-consuming and labor-intensive organic chemistry processes limiting peptide and other biopolymer production. Our goal is to speed drug developers ability to translate therapeutic innovations into clinical impact, said Mytide co-founder Dale Thomas. Our platform takes a holistic view of the entire manufacturing process and couples it with a fully closed-loop computational biology platform, unlocking therapeutic development at unprecedented speeds and precision. The investment from Alloy Therapeutics brings our quick-turn manufacturing technology into a broad drug discovery ecosystem to further accelerate the development of new peptide therapeutics.

Peptides are a high growth drug discovery modality of interest within the pharma industry, with multiple PPVs in Phase III clinical trials. To validate its technology, Mytide has actively partnered its continuous manufacturing platform with pharmaceutical companies requiring scalable and time-sensitive manufacturing for both research and clinical programs. Mytides Gen2 platform is designed to easily be integrated into cGMP manufacturing environments to allow for scalable and decentralized clinical trial manufacturing of a partners lead peptide-based therapeutic candidates. Mytide continues to advance upon the progress in molecular access and analysis being made by the likes of Integrated DNA Technologies (IDT), Illumina, and Thermo Fisher Scientific.

Mytide represents an exciting opportunity to bring down barriers in drug development further, by providing Alloys ecosystem of industry partners with access to high-quality, AI-enabled peptide manufacturing, said Errik Anderson, Alloy Therapeutics CEO and founder. Together, we are excited to empower developers of peptide and combination therapeutics and enable rapid innovation in this promising modality, for the ultimate benefit of patients.

About Mytide Therapeutics:

Mytide Therapeutics is a Boston, MA based peptide and biopolymer manufacturing and computational biology company focused on eliminating the time-consuming and labor-intensive chemical and screening processes preventing innovative therapeutics translation into the clinic. Mytides quick-turn manufacturing technology, coupled with AI-enabled predictive analytics, is providing access to a novel peptide space of difficult-to-make natural and non-natural peptide and peptide conjugates discovery and therapeutic manufacturing. The company is focused on the translation of life-saving therapeutics for serious conditions ranging from metabolic conditions to oncology to inflammatory disorders to infectious diseases.

Learn more about Mytide Therapeutics by visiting Mytide.io or following Mytide on LinkedIn.

About Alloy Therapeutics

Alloy Therapeutics is a biotechnology ecosystem company empowering the global scientific community to make better medicines together. Through a community of partners across academia, biotech, and the largest biopharma, Alloy democratizes access to tools, technologies, services, and company creation capabilities that are foundational for discovering and developing therapeutic biologics. Alloys foundational technology, the ATX-Gx, is a human therapeutic antibody discovery platform consisting of a growing suite of proprietary transgenic mice strains. Alloy is a leader in bispecific antibody discovery and engineering services, utilizing its proprietary ATX-CLC common light chain platform integrating novel transgenic mice and phage display. DeepImmune integrates Alloy's full complement of proprietary in vivo, in vitro, and in silico discovery and optimization technologies into one comprehensive offering for fully human antibody, bispecific, and TCR discovery. DeepImmune is also available for royalty-free access as part of Alloys novel Innovation Subscription model. Alloy is headquartered in Boston, MA with labs in Cambridge, UK; Basel, CH; San Francisco, CA; and Athens, GA. As a reflection of Alloys relentless commitment to the scientific community, Alloy reinvests 100% of its revenue in innovation and access to innovation.

Join the Alloy Therapeutics community by visiting alloytx.com, following Alloy on LinkedIn, or scheduling a 15-minute chat with Alloys Founder and CEO at alloytx.com/ceo.

Originally posted here:
Mytide Therapeutics Raises $7 Million Series A Round to Transform Peptide Manufacturing with Machine Learning - Business Wire