[{"data":1,"prerenderedAt":399},["ShallowReactive",2],{"slug-machine-learning-and-deep-learning-in-medicine":3},{"post":4,"relatedPosts":183,"relatedBooks":316},{"id":5,"date":6,"date_gmt":7,"guid":8,"modified":10,"modified_gmt":11,"slug":12,"status":13,"type":14,"link":15,"title":16,"content":18,"excerpt":21,"author":23,"featured_media":24,"comment_status":25,"ping_status":25,"sticky":26,"template":27,"format":28,"meta":29,"categories":30,"tags":33,"project_category":49,"contact_email_category":50,"yst_prominent_words":51,"class_list":57,"better_featured_image":81,"acf":118,"yoast_meta":128,"_links":130},24020,"2023-01-26T10:00:00","2023-01-26T09:00:00",{"rendered":9},"https:\u002F\u002Fmedicalfuturist.com\u002F?post_id=24020&_wpnonce=9f0c1cfaa7&status=auto-draft&type=post","2023-01-23T09:47:56","2023-01-23T08:47:56","machine-learning-and-deep-learning-in-medicine","publish","post","https:\u002F\u002Fmedicalfuturist.com\u002Fmachine-learning-and-deep-learning-in-medicine",{"rendered":17},"Understanding Machine Learning And Deep Learning In Medicine",{"rendered":19,"protected":20},"\n\u003Cp>Algorithms, datasets, machine learning, deep learning, cognitive computing, big data, and artificial intelligence: IT expressions that took over the language of 21st-century healthcare with surprising force. If medical professionals want to get ahead of the curve, they should get familiarised with the basics of AI and have an idea of what medical problems they aim to solve. So, let’s take a closer look at machine learning and deep learning in medicine.\u003C\u002Fp>\n\n\n\n\u003Ch2 class=\"wp-block-heading\">\u003Cstrong>The\nante-room of artificial intelligence\u003C\u002Fstrong>\u003C\u002Fh2>\n\n\n\n\u003Cp>The term “artificial intelligence” might be misleading as due to the overuse of the expression, its meaning started to get inflated. It implies software with cognition and sentience, a far more developed technology than how it’s used most of the time. For example, Facebook announced an AI to \u003Ca rel=\"noreferrer noopener\" href=\"https:\u002F\u002Fwww.engadget.com\u002F2017\u002F03\u002F01\u002Ffacebook-testing-ai-that-helps-spot-suicidal-users\u002F\" target=\"_blank\">detect suicidal thoughts\u003C\u002Fa> posted to its platform, but \u003Ca rel=\"noreferrer noopener\" href=\"https:\u002F\u002Fwww.theatlantic.com\u002Ftechnology\u002Farchive\u002F2017\u002F03\u002Fwhat-is-artificial-intelligence\u002F518547\u002F\" target=\"_blank\">closer inspection revealed\u003C\u002Fa> that the “AI detection” in question was little more than a pattern-matching filter that flagged posts for human community managers.\u003C\u002Fp>\n\n\n\n\u003Cp>This past year has brought \u003Ca rel=\"noreferrer noopener\" href=\"https:\u002F\u002Fmedicalfuturist.com\u002Fgenerative-adversarial-networks-how-can-ai-learn-so-much-while-its-iq-remains-zero\" target=\"_blank\">vast improvement in the field\u003C\u002Fa> – at least this was when the general public learned about \u003Ca rel=\"noreferrer noopener\" href=\"https:\u002F\u002Fmedicalfuturist.com\u002Fdanger-alert-ai-is-writing-content-but-its-lying-let-us-show-you-how\" target=\"_blank\">revolutionary new algorithms\u003C\u002Fa> (like text-to-image DALL-E and \u003Ca rel=\"noreferrer noopener\" href=\"https:\u002F\u002Fmedicalfuturist.com\u002Fhere-is-how-you-get-friendly-with-a-i-before-it-gets-to-the-office\" target=\"_blank\">Midjourney\u003C\u002Fa>, and large language models like \u003Ca rel=\"noreferrer noopener\" href=\"https:\u002F\u002Fmedicalfuturist.com\u002F6-potential-medical-use-cases-for-chatgpt\" target=\"_blank\">ChatGPT\u003C\u002Fa> and \u003Ca rel=\"noreferrer noopener\" href=\"https:\u002F\u002Fmedicalfuturist.com\u002Fmedpalm-new-ai-medical-chatbots-will-soon-be-better-than-waiting-for-a-doctor\" target=\"_blank\">Google’s MedPaLM\u003C\u002Fa>). \u003C\u002Fp>\n\n\n\n\u003Cp>However impressive these algorithms are, their cognitive capacities still stay below the average human’s. This is what we call \u003Ca rel=\"noreferrer noopener\" href=\"https:\u002F\u002Fwww.sciencealert.com\u002Fnot-everything-we-call-an-ai-is-actually-artificial-intelligence-heres-what-to-know\" target=\"_blank\">artificial narrow intelligence\u003C\u002Fa> (ANI), and the most advanced areas are computer vision and \u003Ca rel=\"noreferrer noopener\" href=\"https:\u002F\u002Fwww.topbots.com\u002Fcategory\u002Ftech\u002Fnlp\u002F\" target=\"_blank\">natural language processing\u003C\u002Fa>. \u003C\u002Fp>\n\n\n\n\u003Cfigure class=\"wp-block-image size-large is-style-default\">\u003Cimg loading=\"lazy\" decoding=\"async\" width=\"768\" height=\"473\" src=\"https:\u002F\u002Fcdn.medicalfuturist.com\u002Fwp-content\u002Fuploads\u002F2023\u002F01\u002FDALL·E-2-768x473.png\" alt=\"\" class=\"wp-image-48811\" srcset=\"https:\u002F\u002Fcdn.medicalfuturist.com\u002Fwp-content\u002Fuploads\u002F2023\u002F01\u002FDALL·E-2-768x473.png 768w, https:\u002F\u002Fcdn.medicalfuturist.com\u002Fwp-content\u002Fuploads\u002F2023\u002F01\u002FDALL·E-2.png 1280w\" sizes=\"auto, (max-width: 768px) 100vw, 768px\" \u002F>\u003Cfigcaption class=\"wp-element-caption\">AI doctors in a hospital – image credit: DALL-E\u003C\u002Ffigcaption>\u003C\u002Ffigure>\n\n\n\n\u003Cp>In very simple terms, ANI has incredible pattern recognition abilities in huge data sets, which makes it perfect for solving text, voice, or image-based classification and clustering problems. \u003C\u002Fp>\n\n\n\n\u003Ch2 class=\"wp-block-heading\">Sometimes you can arrive at great results by using less advanced techniques, too\u003C\u002Fh2>\n\n\n\n\u003Cp>For example,\u003Ca rel=\"noreferrer noopener\" href=\"http:\u002F\u002Fmedicalfuturist.com\u002Fonkonetwork-drastically-reduces-pre-treatment-waiting-time-oncology-patients\u002F\" target=\"_blank\"> in a small Hungarian hospital, the pre-treatment waiting time for oncology patients dropped drastically from 54 to 21 days\u003C\u002Fa> only by optimising patient management processes with the help of simple mechanisms such as recording and following-up cases closely. The first step for data analytics is recording data appropriately – and then carrying out the necessary follow-up actions. The second step is using various statistical methods, such as data mining\u003Ca rel=\"noreferrer noopener\" href=\"https:\u002F\u002Fwww.topbots.com\u002Fwtf-is-artificial-intelligence-intro\u002F\" target=\"_blank\"> for collecting, analysing, describing, visualising and drawing inferences from data\u003C\u002Fa>, for example from electronic health records or \u003Ca rel=\"noreferrer noopener\" href=\"https:\u002F\u002Fmedicalfuturist.com\u002Fthe-omics-universe-and-its-future\" target=\"_blank\">the ‘OMICS’ universe\u003C\u002Fa>. The focus is on discovering mathematical relationships and properties within big data sets and quantifying uncertainty. Data mining usually means when\u003Ca rel=\"noreferrer noopener\" href=\"https:\u002F\u002Fwww.topbots.com\u002Fwtf-is-artificial-intelligence-intro\u002F\" target=\"_blank\"> insights and patterns are extracted from large-scale databases\u003C\u002Fa>. \u003C\u002Fp>\n\n\n\n\u003Cp>However, this is only the ante-room of artificial intelligence: machine learning and deep learning go far beyond that.\u003C\u002Fp>\n\n\n\n\u003Ch2 class=\"wp-block-heading\">Pattern recognizers rule the world: supervised machine learning and measles\u003C\u002Fh2>\n\n\n\n\u003Cp>Machine learning is the field of computer science that\u003Ca rel=\"noreferrer noopener\" href=\"https:\u002F\u002Fwww.topbots.com\u002Fwtf-is-artificial-intelligence-intro\u002F\" target=\"_blank\"> enables computers to learn without being explicitly programmed\u003C\u002Fa> building on top of computational statistics and data mining. As with traditional statistics, \u003Ca rel=\"noreferrer noopener\" href=\"https:\u002F\u002Fwww.sciencedirect.com\u002Fscience\u002Farticle\u002Fpii\u002FS0735109717368456\" target=\"_blank\">machine learning requires sufficient training datasets\u003C\u002Fa> (also known as sample size in traditional statistics) and the right algorithms to optimise its performance on the training dataset before testing. However, in contrast to traditional methods, machine learning is focused on building automated decision systems.\u003C\u002Fp>\n\n\n\n\u003Cp>The field has many different types: it could be\u003Ca rel=\"noreferrer noopener\" href=\"https:\u002F\u002Fwww.topbots.com\u002Fwtf-is-artificial-intelligence-intro\u002F\" target=\"_blank\"> supervised, unsupervised, semi-supervised or reinforcement learning\u003C\u002Fa>, among others. And many algorithms are in fact “hybrids”, for example, ChatGPT was trained with supervised and reinforcement learning.\u003C\u002Fp>\n\n\n\n\u003Cp>\u003Cstrong>Supervised learning\u003C\u002Fstrong> is typically used for classification problems, e.g. pairing pictures with labels. You have input and output data – the image as well as the label -; and the algorithm learns general rules on how to categorise. It is the most popular type of machine learning in medicine, and in a few years, it will be widely used in medical imaging in \u003Ca rel=\"noreferrer noopener\" href=\"https:\u002F\u002Fmedicalfuturist.com\u002Fthe-future-of-radiology-and-ai\" target=\"_blank\">radiology\u003C\u002Fa>, \u003Ca rel=\"noreferrer noopener\" href=\"https:\u002F\u002Fmedicalfuturist.com\u002Fdigital-future-pathology\" target=\"_blank\">pathology\u003C\u002Fa>, and other image-intensive fields. Although it certainly has its limitations: it requires large data sets to become accurate enough, and the data has to be appropriately labelled. That’s why the \u003Ca rel=\"noreferrer noopener\" href=\"https:\u002F\u002Fmedicalfuturist.com\u002Fdata-annotation\" target=\"_blank\">work of data annotators is so crucial\u003C\u002Fa>.\u003C\u002Fp>\n\n\n\n\u003Cp>Nevertheless, supervised machine learning can also be effectively deployed to predict health events based on various input data, which can be classified in a linear way. For example, \u003Ca rel=\"noreferrer noopener\" href=\"https:\u002F\u002Fwww.medicaldaily.com\u002Fhow-big-data-and-machine-learning-can-predict-prevent-isolated-cases-disease-435359\" target=\"_blank\">from statistics on measles vaccination rates and disease outbreaks from the Centers for Disease Control and Prevention\u003C\u002Fa>, as well as non-traditional health data, including social media and syndromic surveillance data generated by software that mines a huge range of medical records sources, an algorithm can provide a reliable map of future measles outbreak hotspots.\u003C\u002Fp>\n\n\n\n\u003Cfigure class=\"wp-block-image size-large is-style-default\">\u003Cimg loading=\"lazy\" decoding=\"async\" width=\"768\" height=\"492\" src=\"https:\u002F\u002Fcdn.medicalfuturist.com\u002Fwp-content\u002Fuploads\u002F2023\u002F01\u002FDALL·E-pandemic-outbreak-world-map-digital-art-768x492.png\" alt=\"\" class=\"wp-image-48813\" srcset=\"https:\u002F\u002Fcdn.medicalfuturist.com\u002Fwp-content\u002Fuploads\u002F2023\u002F01\u002FDALL·E-pandemic-outbreak-world-map-digital-art-768x492.png 768w, https:\u002F\u002Fcdn.medicalfuturist.com\u002Fwp-content\u002Fuploads\u002F2023\u002F01\u002FDALL·E-pandemic-outbreak-world-map-digital-art-1536x983.png 1536w, https:\u002F\u002Fcdn.medicalfuturist.com\u002Fwp-content\u002Fuploads\u002F2023\u002F01\u002FDALL·E-pandemic-outbreak-world-map-digital-art.png 1600w\" sizes=\"auto, (max-width: 768px) 100vw, 768px\" \u002F>\u003Cfigcaption class=\"wp-element-caption\">Pandemic outbreak world map – image credit: DALL-E\u003C\u002Ffigcaption>\u003C\u002Ffigure>\n\n\n\n\u003Ch2 class=\"wp-block-heading\">\u003Cstrong>Unsupervised\nmachine learning and drug interactions\u003C\u002Fstrong>\u003C\u002Fh2>\n\n\n\n\u003Cp>In the case of \u003Cstrong>unsupervised machine learning\u003C\u002Fstrong>, the computer program is asked to discover inherent structures and patterns that lie within the data. Unlike in the case of supervised machine learning, the data sets are unlabeled and unstructured. Thus, the algorithm has to make up its own groups, clusters, and categories based on “similarities” in huge data sets. It is usually used \u003Ca rel=\"noreferrer noopener\" href=\"https:\u002F\u002Fwww.sciencedirect.com\u002Fscience\u002Farticle\u002Fpii\u002FS0735109717368456\" target=\"_blank\">to predict unknown results and to determine how to discover hidden patterns\u003C\u002Fa>. Unsupervised machine learning has subtypes: clustering algorithms and association rule-learning algorithms.\u003C\u002Fp>\n\n\n\n\u003Cp>\u003Ca rel=\"noreferrer noopener\" href=\"https:\u002F\u002Fmedicalfuturist.com\u002Fgenerative-adversarial-networks-how-can-ai-learn-so-much-while-its-iq-remains-zero\u002F\" target=\"_blank\">Text-to-image algorithm Midjourney\u003C\u002Fa>, which is a Generative Adversarial Network (GAN), also learns by unsupervised learning, although there is a twist: it actually has two algorithms that teach each other in a zero-sum game, and this results in us, users getting better and better images created by the AI from scratch. \u003C\u002Fp>\n\n\n\n\u003Cp>Unsupervised learning is often used in deep learning, and has been implemented in self-driving vehicles and robots as well as being used in speech- and pattern-recognition applications. In medicine, for example, \u003Ca rel=\"noreferrer noopener\" href=\"https:\u002F\u002Fsubscription.packtpub.com\u002Fbook\u002Fbig_data_and_business_intelligence\u002F9781785880513\u002F1\u002Fch01lvl1sec11\u002Fmachine-learning-types-and-subtypes\" target=\"_blank\">tissue samples can be clustered based on similar gene expression values\u003C\u002Fa> using unsupervised learning techniques. As an example of association rule-learning algorithms, the \u003Ca rel=\"noreferrer noopener\" href=\"https:\u002F\u002Fwww.sciencedirect.com\u002Fscience\u002Farticle\u002Fpii\u002FS0735109717368456\" target=\"_blank\">testing of novel drug-drug interactions\u003C\u002Fa> can be mentioned.\u003C\u002Fp>\n\n\n\n\u003Ch2 class=\"wp-block-heading\">\u003Cstrong>Reinforcement\nlearning and the magic of AlphaGo\u003C\u002Fstrong>\u003C\u002Fh2>\n\n\n\n\u003Cp>The last category, \u003Cstrong>reinforcement learning\u003C\u002Fstrong> constitutes probably the most known type of machine learning: when the computer program learns from its mistakes and successes and builds its experiences into the algorithm. A well-known example is \u003Ca rel=\"noreferrer noopener\" href=\"https:\u002F\u002Fmedicalfuturist.com\u002Fartificial-intelligence-and-the-art-of-medicine\" target=\"_blank\">AlphaGo, the machine developed by Google that decisively beat the World Go Champion Lee Sedol in March 2016\u003C\u002Fa>. Using a reward and penalty scheme, the model first trained on millions of board positions in the supervised learning stage, then played itself in the reinforcement learning stage to ultimately become good enough to triumph over the best human player.\u003C\u002Fp>\n\n\n\n\u003Cp>However, the problem with applying reinforcement learning to healthcare, especially for optimising treatment, is that \u003Ca rel=\"noreferrer noopener\" href=\"https:\u002F\u002Ftowardsdatascience.com\u002Fa-review-of-recent-reinforcment-learning-applications-to-healthcare-1f8357600407\" target=\"_blank\">unlike with AlphaGo, we cannot play out a large number of scenarios where the agent makes interventions to learn the optimal policy\u003C\u002Fa> – as the lives of patients are at stake. Luckily, we already have examples where this issue can be mitigated. \u003C\u002Fp>\n\n\n\n\u003Cp>In a \u003Ca rel=\"noreferrer noopener\" href=\"http:\u002F\u002Fweb.media.mit.edu\u002F~pratiks\u002Fmlhc_2018\u002Freinforcement_learning_with_action_derived_rewards_for_chemotherapy_and_clinical_trial_dosing_regimen_selection.pdf\" target=\"_blank\">study published by MIT researchers\u003C\u002Fa>, the authors reported a successful formulation of clinical trial dosing as a reinforcement learning problem, where the algorithm taught the appropriate dosing regiments to reduce mean tumor diameters in patients undergoing chemo- and radiation therapy clinical trials. \u003C\u002Fp>\n\n\n\n\u003Cfigure class=\"wp-block-image size-large is-style-default\">\u003Cimg loading=\"lazy\" decoding=\"async\" width=\"768\" height=\"512\" src=\"https:\u002F\u002Fcdn.medicalfuturist.com\u002Fwp-content\u002Fuploads\u002F2023\u002F01\u002FMidjourney_supercomputer_and_human_playing_chess-768x512.png\" alt=\"Midjourney Supercomputer and human playing chess AI algorithm\" class=\"wp-image-48815\" srcset=\"https:\u002F\u002Fcdn.medicalfuturist.com\u002Fwp-content\u002Fuploads\u002F2023\u002F01\u002FMidjourney_supercomputer_and_human_playing_chess-768x512.png 768w, https:\u002F\u002Fcdn.medicalfuturist.com\u002Fwp-content\u002Fuploads\u002F2023\u002F01\u002FMidjourney_supercomputer_and_human_playing_chess.png 1280w\" sizes=\"auto, (max-width: 768px) 100vw, 768px\" \u002F>\u003Cfigcaption class=\"wp-element-caption\">Supercomputer and human playing chess – image credit – Midjourney\u003C\u002Ffigcaption>\u003C\u002Ffigure>\n\n\n\n\u003Ch2 class=\"wp-block-heading\">\u003Cstrong>Deep\nlearning in medicine used for very complex issues\u003C\u002Fstrong>\u003C\u002Fh2>\n\n\n\n\u003Cp>\u003Cstrong>Deep learning\u003C\u002Fstrong> is the subfield of machine learning where computers learn with the help of layered neural networks. However, there’s no strict line between machine and deep learning; usually, the cleanliness of data and the complexity of the problem determine which one is more applicable. Deep learning algorithms usually deal with messy datasets, and unstructured piles of information to try to give answers to difficult questions.\u003C\u002Fp>\n\n\n\n\u003Cp>What are \u003Cstrong>neural networks\u003C\u002Fstrong>? Their operation basically imitates the neurons in the brain. \u003Ca rel=\"noreferrer noopener\" href=\"https:\u002F\u002Fqz.com\u002F1046350\u002Fthe-quartz-guide-to-artificial-intelligence-what-is-it-why-is-it-important-and-should-we-be-afraid\u002F\" target=\"_blank\">Quartz formulated the explanation as the followings\u003C\u002Fa>: algorithms which are roughly built to model the way the brain processes information, through webs of connected mathematic equations. Data given to a neural network is broken into smaller pieces and analysed for underlying patterns thousands to millions of times depending on the complexity of the network. A deep neural network is when the output of one neural network is fed into the input of another, chaining them together as layers. Typically, the layers of a deep neural network would analyse data on higher and higher levels of abstraction, meaning they each throw out data learned to be unnecessary until the simplest and most accurate representation of the data is left.\u003C\u002Fp>\n\n\n\n\u003Cp>Deep learning has different types based on the ways of connecting layers and the ways ‘neurons’ act. There’s also unsupervised, supervised and reinforcement learning in deep learning algorithms, as these signify the way the algorithm is fed with data by researchers. \u003C\u002Fp>\n\n\n\n\u003Cp>Beyond these, convolutional neural networks (CNN) are typical for recognising images, video, and audio data, due to their ability to work with dense data. Recurrent neural networks (RNN) are used for natural language processing, while long short-term memory networks (LSTM) are variations of RNNs meant to retain structured information based on data. For instance, an RNN could recognise all the nouns and adjectives in a sentence and determine if they’re used correctly, and an LSTM could remember the plot of a book.\u003C\u002Fp>\n\n\n\n\u003Cp>As an example of deep learning in medicine, \u003Ca href=\"https:\u002F\u002Farxiv.org\u002Fabs\u002F1711.09602\" target=\"_blank\" rel=\"noreferrer noopener\">researchers proposed an approach to deduce treatment policies for septic patients\u003C\u002Fa> by using continuous state-space models and deep reinforcement learning. In \u003Ca href=\"https:\u002F\u002Flink.springer.com\u002Fchapter\u002F10.1007\u002F978-3-030-00934-2_68\" target=\"_blank\" rel=\"noreferrer noopener\">another study\u003C\u002Fa>, experts attempt to solve the difficult problem of estimating polyp size using colonoscopy images or videos, which is crucial for making a diagnosis in colon cancer screening. Moreover, \u003Ca href=\"https:\u002F\u002Fwww.sciencedirect.com\u002Fscience\u002Farticle\u002Fpii\u002FS0735109717368456\" target=\"_blank\" rel=\"noreferrer noopener\">unsupervised deep learning may facilitate the exploration of novel factors in score systems or add hidden risk factors\u003C\u002Fa> to existing models. It can also be used to classify novel genotypes and phenotypes from pulmonary hypertension, cardiomyopathy, and many other factors.\u003C\u002Fp>\n\n\n\n\u003Cp>Navigating the sea of information about artificial intelligence is tough. As everyone realized that the technologies leading to AI could revolutionise healthcare, there are a lot of experiments and research, but also bogus information and overhype out there. We, at The Medical Futurist, aim to provide you with context to interpret study results and attempt to make sense of the digital health revolution. Feel free to reach out to us for questions, comments or just a conversation. We’d love to hear from you!\u003C\u002Fp>\n",false,{"rendered":22,"protected":20},"\u003Cp>Algorithms, datasets, machine learning, deep learning, cognitive computing, big data, and artificial intelligence: IT expressions that took over the language of 21st-century healthcare with surprising force. If medical professionals want to get ahead of the curve, they rather get familiarized with the basics of A.I. and have an idea of what medical problems they aim to solve. So, let’s take a closer look at machine learning and deep learning in medicine.\u003C\u002Fp>\n",6,24025,"closed",true,"","standard",{"_acf_changed":20,"footnotes":27},[31,32],504,521,[34,35,36,37,38,39,40,41,42,43,44,45,46,47,48],137,671,144,1168,246,1203,271,1228,275,1350,289,313,425,134,636,[],[],[52,53,54,55,56],1693,1715,1789,2695,2739,[58,14,59,60,61,62,63,64,65,66,67,68,69,70,71,72,73,74,75,76,77,78,79,80],"post-24020","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-artificial-intelligence","category-future-medicine","tag-algorithm","tag-machine-learning","tag-artificial-intelligence","tag-smart-algorithm","tag-future","tag-smart","tag-health","tag-artificial","tag-healthcare","tag-smart-health","tag-innovation","tag-medicine","tag-technology-2","tag-ai","tag-deep-learning",{"id":24,"alt_text":82,"caption":27,"description":27,"media_type":83,"media_details":84,"post":5,"source_url":117},"machine learning and deep learning in medicine","image",{"width":85,"height":86,"file":87,"sizes":88,"image_meta":115},1920,1080,"2019\u002F05\u002F088_mlearning_dlearning.png",{"medium":89,"large":95,"thumbnail":100,"medium_large":104,"1536x1536":105,"2048x2048":110},{"file":90,"width":91,"height":92,"mime-type":93,"source_url":94},"088_mlearning_dlearning-370x208.png","370","208","image\u002Fpng","https:\u002F\u002Fcdn.medicalfuturist.com\u002Fwp-content\u002Fuploads\u002F2019\u002F05\u002F088_mlearning_dlearning-370x208.png",{"file":96,"width":97,"height":98,"mime-type":93,"source_url":99},"088_mlearning_dlearning-768x432.png","768","432","https:\u002F\u002Fcdn.medicalfuturist.com\u002Fwp-content\u002Fuploads\u002F2019\u002F05\u002F088_mlearning_dlearning-768x432.png",{"file":101,"width":102,"height":102,"mime-type":93,"source_url":103},"088_mlearning_dlearning-150x150.png","150","https:\u002F\u002Fcdn.medicalfuturist.com\u002Fwp-content\u002Fuploads\u002F2019\u002F05\u002F088_mlearning_dlearning-150x150.png",{"file":96,"width":97,"height":98,"mime-type":93,"source_url":99},{"file":106,"width":107,"height":108,"mime-type":93,"source_url":109},"088_mlearning_dlearning-1536x864.png","1536","864","https:\u002F\u002Fcdn.medicalfuturist.com\u002Fwp-content\u002Fuploads\u002F2019\u002F05\u002F088_mlearning_dlearning-1536x864.png",{"file":111,"width":112,"height":113,"mime-type":93,"source_url":114},"088_mlearning_dlearning-2048x1152.png","2048","1152","https:\u002F\u002Fcdn.medicalfuturist.com\u002Fwp-content\u002Fuploads\u002F2019\u002F05\u002F088_mlearning_dlearning-2048x1152.png",{"aperture":116,"credit":27,"camera":27,"caption":27,"created_timestamp":116,"copyright":27,"focal_length":116,"iso":116,"shutter_speed":116,"title":27,"orientation":116},"0","https:\u002F\u002Fcdn.medicalfuturist.com\u002Fwp-content\u002Fuploads\u002F2019\u002F05\u002F088_mlearning_dlearning-scaled.png",{"related_posts":20,"related_posts_footer":119,"cta_type":123,"cta_color":27,"related_books":124,"subtitle":27},[120,121,122],47975,48505,48665,"subscribe",[125,126,127],24762,47427,37033,{"yoast_wpseo_title":17,"yoast_wpseo_metadesc":129,"yoast_wpseo_canonical":15},"Here, we take a closer look at machine learning and deep learning in medicine, focusing especially on the real-life problems these can solve in healthcare.",{"self":131,"collection":137,"about":140,"author":143,"replies":146,"version-history":149,"predecessor-version":153,"wp:featuredmedia":157,"wp:attachment":160,"wp:term":163,"curies":179},[132],{"href":133,"targetHints":134},"https:\u002F\u002Fapi.medicalfuturist.com\u002Fwp-json\u002Fwp\u002Fv2\u002Fposts\u002F24020",{"allow":135},[136],"GET",[138],{"href":139},"https:\u002F\u002Fapi.medicalfuturist.com\u002Fwp-json\u002Fwp\u002Fv2\u002Fposts",[141],{"href":142},"https:\u002F\u002Fapi.medicalfuturist.com\u002Fwp-json\u002Fwp\u002Fv2\u002Ftypes\u002Fpost",[144],{"embeddable":26,"href":145},"https:\u002F\u002Fapi.medicalfuturist.com\u002Fwp-json\u002Fwp\u002Fv2\u002Fusers\u002F6",[147],{"embeddable":26,"href":148},"https:\u002F\u002Fapi.medicalfuturist.com\u002Fwp-json\u002Fwp\u002Fv2\u002Fcomments?post=24020",[150],{"count":151,"href":152},26,"https:\u002F\u002Fapi.medicalfuturist.com\u002Fwp-json\u002Fwp\u002Fv2\u002Fposts\u002F24020\u002Frevisions",[154],{"id":155,"href":156},48875,"https:\u002F\u002Fapi.medicalfuturist.com\u002Fwp-json\u002Fwp\u002Fv2\u002Fposts\u002F24020\u002Frevisions\u002F48875",[158],{"embeddable":26,"href":159},"https:\u002F\u002Fapi.medicalfuturist.com\u002Fwp-json\u002Fwp\u002Fv2\u002Fmedia\u002F24025",[161],{"href":162},"https:\u002F\u002Fapi.medicalfuturist.com\u002Fwp-json\u002Fwp\u002Fv2\u002Fmedia?parent=24020",[164,167,170,173,176],{"taxonomy":165,"embeddable":26,"href":166},"category","https:\u002F\u002Fapi.medicalfuturist.com\u002Fwp-json\u002Fwp\u002Fv2\u002Fcategories?post=24020",{"taxonomy":168,"embeddable":26,"href":169},"post_tag","https:\u002F\u002Fapi.medicalfuturist.com\u002Fwp-json\u002Fwp\u002Fv2\u002Ftags?post=24020",{"taxonomy":171,"embeddable":26,"href":172},"project_category","https:\u002F\u002Fapi.medicalfuturist.com\u002Fwp-json\u002Fwp\u002Fv2\u002Fproject_category?post=24020",{"taxonomy":174,"embeddable":26,"href":175},"contact_email_category","https:\u002F\u002Fapi.medicalfuturist.com\u002Fwp-json\u002Fwp\u002Fv2\u002Fcontact_email_category?post=24020",{"taxonomy":177,"embeddable":26,"href":178},"yst_prominent_words","https:\u002F\u002Fapi.medicalfuturist.com\u002Fwp-json\u002Fwp\u002Fv2\u002Fyst_prominent_words?post=24020",[180],{"name":181,"href":182,"templated":26},"wp","https:\u002F\u002Fapi.w.org\u002F{rel}",[184],{"id":122,"date":185,"date_gmt":186,"guid":187,"modified":189,"modified_gmt":190,"slug":191,"status":13,"type":14,"link":192,"title":193,"content":195,"excerpt":197,"author":23,"featured_media":199,"comment_status":25,"ping_status":25,"sticky":26,"template":27,"format":28,"meta":200,"categories":201,"tags":202,"project_category":213,"contact_email_category":215,"yst_prominent_words":216,"class_list":223,"better_featured_image":236,"acf":265,"yoast_meta":271,"_links":274},"2023-01-17T10:00:00","2023-01-17T09:00:00",{"rendered":188},"https:\u002F\u002Fmedicalfuturist.com\u002F?post_id=48665&_wpnonce=dff5872b54&status=auto-draft&type=post","2023-01-17T07:42:35","2023-01-17T06:42:35","medpalm-new-ai-medical-chatbots-will-soon-be-better-than-waiting-for-a-doctor","https:\u002F\u002Fmedicalfuturist.com\u002Fmedpalm-new-ai-medical-chatbots-will-soon-be-better-than-waiting-for-a-doctor",{"rendered":194},"MedPaLM: New AI Medical Chatbots Will Soon Be Better Than Waiting For A Doctor",{"rendered":196,"protected":20},"\n\u003Cp>Large language models (LLMs), these excitingly versatile algorithms became a topic of general conversation in December 2022, when OpenAI released its GPT3 agent, also known as ChatGPT. LLMs are developed to carry on conversations in human-like ways, they are designed to understand complex queries and respond in a nuanced manner. We introduced potential medical use cases \u003Ca href=\"https:\u002F\u002Fmedicalfuturist.com\u002F6-potential-medical-use-cases-for-chatgpt\" target=\"_blank\" rel=\"noreferrer noopener\">in this article\u003C\u002Fa>. \u003C\u002Fp>\n\n\n\n\u003Cp>By now we all have a track record with some kind of a chatbot – if nothing else, a not-very-bright algorithm at a service provider. These interactions rarely left anyone particularly impressed, they sometimes contribute to solving our problems, but more often than not just result in a frustrated user leaving with a promise of contact from a human support staff member that may or may never happen. \u003C\u002Fp>\n\n\n\n\u003Cp>Large language models undoubtedly have changed this field forever, they are capable of such high-quality assistance that was never seen earlier. Only a few short weeks after the release of ChatGPT, Google\u002FDeepMind announced \u003Ca href=\"https:\u002F\u002Fpharmaphorum.com\u002Fnews\u002Fgoogle-and-deepmind-share-work-on-medical-chatbot-med-palm\u002F\" target=\"_blank\" rel=\"noreferrer noopener\">the release of MedPaLM\u003C\u002Fa>, a large language model specifically designed to answer healthcare-related questions, based on their 540-billion parameter PaLM model.\u003C\u002Fp>\n\n\n\n\u003Cfigure class=\"wp-block-image size-large is-style-default\">\u003Cimg loading=\"lazy\" decoding=\"async\" width=\"768\" height=\"351\" src=\"https:\u002F\u002Fcdn.medicalfuturist.com\u002Fwp-content\u002Fuploads\u002F2023\u002F01\u002FMedpalm2-768x351.jpg\" alt=\"Medpalm google deepmind large languae model\" class=\"wp-image-48673\" srcset=\"https:\u002F\u002Fcdn.medicalfuturist.com\u002Fwp-content\u002Fuploads\u002F2023\u002F01\u002FMedpalm2-768x351.jpg 768w, https:\u002F\u002Fcdn.medicalfuturist.com\u002Fwp-content\u002Fuploads\u002F2023\u002F01\u002FMedpalm2.jpg 988w\" sizes=\"auto, (max-width: 768px) 100vw, 768px\" \u002F>\u003C\u002Ffigure>\n\n\n\n\u003Cp>This model was trained on six existing medical Q&A datasets (NedQA, MedMCQA, PubMedQA, LiveQA, MedicationQA, and MMLU), and the developer teams also created their own HealthSearchQA, using questions about medical conditions and the associated symptoms.\u003C\u002Fp>\n\n\n\n\u003Cp>At the moment MedPaLM can’t be tested by the general public, but you can read the \u003Ca href=\"https:\u002F\u002Farxiv.org\u002Fpdf\u002F2212.13138.pdf\" target=\"_blank\" rel=\"noreferrer noopener\">researcher’s paper here\u003C\u002Fa>. \u003C\u002Fp>\n\n\n\n\u003Ch3 class=\"wp-block-heading\">It “performs encouragingly but remains inferior to clinicians”\u003C\u002Fh3>\n\n\n\n\u003Cp>The document lists a number of possible medical applications, including knowledge retrieval, clinical decision support, summarisation of key findings in studies, and triaging patients’ primary care concerns among others, but also noted that MedPaLM “performs encouragingly, but remains inferior to clinicians.”\u003C\u002Fp>\n\n\n\n\u003Cp>The attached diagrams show that MedPaLM was still underperforming human clinicians in several areas: \u003C\u002Fp>\n\n\n\n\u003Cul class=\"wp-block-list\">\n\u003Cli>incorrect retrieval of information was 16.9% for Med-PaLM, which compares to 3.6% of clinicians\u003C\u002Fli>\n\n\n\n\u003Cli>incorrect reasoning was seen in 10.1% of the MedPaLM answers and in 2.1% of clinician answers\u003C\u002Fli>\n\n\n\n\u003Cli>incorrect comprehension happened in 18.7% of cases for the algorithm and in 2.2% for the clinicians\u003C\u002Fli>\n\u003C\u002Ful>\n\n\n\n\u003Cfigure class=\"wp-block-image size-large is-style-default\">\u003Cimg loading=\"lazy\" decoding=\"async\" width=\"768\" height=\"533\" src=\"https:\u002F\u002Fcdn.medicalfuturist.com\u002Fwp-content\u002Fuploads\u002F2023\u002F01\u002Fmedpalm-768x533.jpg\" alt=\"\" class=\"wp-image-48669\" srcset=\"https:\u002F\u002Fcdn.medicalfuturist.com\u002Fwp-content\u002Fuploads\u002F2023\u002F01\u002Fmedpalm-768x533.jpg 768w, https:\u002F\u002Fcdn.medicalfuturist.com\u002Fwp-content\u002Fuploads\u002F2023\u002F01\u002Fmedpalm.jpg 995w\" sizes=\"auto, (max-width: 768px) 100vw, 768px\" \u002F>\u003Cfigcaption class=\"wp-element-caption\">Some examples of answers crafted by MedPaLM\u003C\u002Ffigcaption>\u003C\u002Ffigure>\n\n\n\n\u003Cp>\u003Cstrong>Large language models may easily be the best option we’ll have for medical consultations\u003C\u002Fstrong>\u003C\u002Fp>\n\n\n\n\u003Cp>Although the model is obviously not perfect, it does significantly better than any previous algorithms, and the field is improving fast. The LLM-based chatbot algorithms provide a never-before-seen quality of human-AI interaction.\u003C\u002Fp>\n\n\n\n\u003Cp>This is what I think will happen: as these models get better and better, the risk of missing care due to capacity shortages in healthcare will soon outweigh the risk of the algorithms being wrong. We will be better off familiarising ourselves with communicating with such an LLM algorithm – purely because long waiting for medical answers due to the lack of healthcare personnel will pose a higher threat. \u003C\u002Fp>\n\n\n\n\u003Cp>Live consultation with a doctor will become a luxury in the 21st century, and any solution addressing this issue (from asynchronous telemedicine to medical chatbots) will actually improve our health prospects.\u003C\u002Fp>\n",{"rendered":198,"protected":20},"\u003Cp>As these models get better and better, the risk of missing care due to capacity shortages in healthcare will soon outweigh the risk of the algorithms being wrong. \u003C\u002Fp>\n",48671,{"_acf_changed":20,"footnotes":27},[31],[203,204,205,206,207,208,209,210,211,212],734,7739,7741,7743,7745,7747,7749,7687,7691,7737,[214],950,[],[217,218,219,220,221,222],1723,2313,2649,2689,2693,2705,[224,14,59,60,61,62,63,64,225,226,227,228,229,230,231,232,233,234,235],"post-48665","tag-chatbots","tag-ai-chatbots","tag-large-language-models","tag-chat-gpt","tag-medpalm","tag-openai","tag-deep-mind","tag-ai-text-generator","tag-ai-in-heaalthcare","tag-medical-chatbots","project_category-medical-professionals",{"id":199,"alt_text":237,"caption":27,"description":27,"media_type":83,"media_details":238,"post":122,"source_url":264},"medical chatbot AI algorithm person man phone TMF",{"width":239,"height":240,"file":241,"filesize":242,"sizes":243,"image_meta":262},6667,3750,"2023\u002F01\u002Ftmf_article_348-01.png",1050118,{"medium":244,"large":250,"thumbnail":256,"medium_large":261},{"file":245,"width":246,"height":247,"mime-type":93,"filesize":248,"source_url":249},"tmf_article_348-01-370x208.png",370,208,23029,"https:\u002F\u002Fcdn.medicalfuturist.com\u002Fwp-content\u002Fuploads\u002F2023\u002F01\u002Ftmf_article_348-01-370x208.png",{"file":251,"width":252,"height":253,"mime-type":93,"filesize":254,"source_url":255},"tmf_article_348-01-768x432.png",768,432,59384,"https:\u002F\u002Fcdn.medicalfuturist.com\u002Fwp-content\u002Fuploads\u002F2023\u002F01\u002Ftmf_article_348-01-768x432.png",{"file":257,"width":258,"height":258,"mime-type":93,"filesize":259,"source_url":260},"tmf_article_348-01-150x150.png",150,11133,"https:\u002F\u002Fcdn.medicalfuturist.com\u002Fwp-content\u002Fuploads\u002F2023\u002F01\u002Ftmf_article_348-01-150x150.png",{"file":251,"width":252,"height":253,"mime-type":93,"filesize":254,"source_url":255},{"aperture":116,"credit":27,"camera":27,"caption":27,"created_timestamp":116,"copyright":27,"focal_length":116,"iso":116,"shutter_speed":116,"title":27,"orientation":116,"keywords":263},[],"https:\u002F\u002Fcdn.medicalfuturist.com\u002Fwp-content\u002Fuploads\u002F2023\u002F01\u002Ftmf_article_348-01.png",{"cta_type":123,"cta_color":27,"related_books":266,"related_posts_footer":269,"related_posts":20,"subtitle":27},[125,267,268],34151,24759,[121,120,270],47785,{"yoast_wpseo_title":272,"yoast_wpseo_metadesc":273,"yoast_wpseo_canonical":192},"MedPaLM: New Chatbots Will Soon Be Better Than Waiting For A Doctor","As models like MedPaLM get better, the risk of missing care due to capacity shortages in healthcare will outweigh the risk of the algorithms being wrong.",{"self":275,"collection":280,"about":282,"author":284,"replies":286,"version-history":289,"predecessor-version":293,"wp:featuredmedia":297,"wp:attachment":300,"wp:term":303,"curies":314},[276],{"href":277,"targetHints":278},"https:\u002F\u002Fapi.medicalfuturist.com\u002Fwp-json\u002Fwp\u002Fv2\u002Fposts\u002F48665",{"allow":279},[136],[281],{"href":139},[283],{"href":142},[285],{"embeddable":26,"href":145},[287],{"embeddable":26,"href":288},"https:\u002F\u002Fapi.medicalfuturist.com\u002Fwp-json\u002Fwp\u002Fv2\u002Fcomments?post=48665",[290],{"count":291,"href":292},7,"https:\u002F\u002Fapi.medicalfuturist.com\u002Fwp-json\u002Fwp\u002Fv2\u002Fposts\u002F48665\u002Frevisions",[294],{"id":295,"href":296},48723,"https:\u002F\u002Fapi.medicalfuturist.com\u002Fwp-json\u002Fwp\u002Fv2\u002Fposts\u002F48665\u002Frevisions\u002F48723",[298],{"embeddable":26,"href":299},"https:\u002F\u002Fapi.medicalfuturist.com\u002Fwp-json\u002Fwp\u002Fv2\u002Fmedia\u002F48671",[301],{"href":302},"https:\u002F\u002Fapi.medicalfuturist.com\u002Fwp-json\u002Fwp\u002Fv2\u002Fmedia?parent=48665",[304,306,308,310,312],{"taxonomy":165,"embeddable":26,"href":305},"https:\u002F\u002Fapi.medicalfuturist.com\u002Fwp-json\u002Fwp\u002Fv2\u002Fcategories?post=48665",{"taxonomy":168,"embeddable":26,"href":307},"https:\u002F\u002Fapi.medicalfuturist.com\u002Fwp-json\u002Fwp\u002Fv2\u002Ftags?post=48665",{"taxonomy":171,"embeddable":26,"href":309},"https:\u002F\u002Fapi.medicalfuturist.com\u002Fwp-json\u002Fwp\u002Fv2\u002Fproject_category?post=48665",{"taxonomy":174,"embeddable":26,"href":311},"https:\u002F\u002Fapi.medicalfuturist.com\u002Fwp-json\u002Fwp\u002Fv2\u002Fcontact_email_category?post=48665",{"taxonomy":177,"embeddable":26,"href":313},"https:\u002F\u002Fapi.medicalfuturist.com\u002Fwp-json\u002Fwp\u002Fv2\u002Fyst_prominent_words?post=48665",[315],{"name":181,"href":182,"templated":26},[317],{"id":127,"date":318,"date_gmt":319,"guid":320,"modified":322,"modified_gmt":323,"slug":324,"status":13,"type":325,"link":326,"title":327,"content":329,"excerpt":331,"author":23,"featured_media":333,"comment_status":25,"ping_status":25,"template":27,"yst_prominent_words":334,"class_list":335,"better_featured_image":338,"acf":356,"yoast_meta":368,"_links":371},"2021-12-02T11:04:38","2021-12-02T10:04:38",{"rendered":321},"https:\u002F\u002Fapi.medicalfuturist.com\u002F?post_type=book&p=37033","2025-05-29T13:51:30","2025-05-29T11:51:30","hype-cycle-of-the-top-50-emerging-digital-health-trends","book","https:\u002F\u002Fapi.medicalfuturist.com\u002Fbooks\u002Fhype-cycle-of-the-top-50-emerging-digital-health-trends\u002F",{"rendered":328},"The Technology Adoption Curve Of The Top 50 Emerging Digital Health Trends",{"rendered":330,"protected":20},"\n\u003Cp>This is the first time we publish The Medical Futurist’s Technology Adoption Curve, showing 50 of the most promising digital health technologies the way we, at The Medical Futurist, see them today. \u003C\u002Fp>\n\n\n\n\u003Cp>\u003C\u002Fp>\n",{"rendered":332,"protected":20},"\u003Cp>This is the first time we publish The Medical Futurist’s Technology Adoption Curve, showing 50 of the most promising digital health technologies the way we, […]\u003C\u002Fp>\n",49955,[217],[336,325,337,60,62,63],"post-37033","type-book",{"id":333,"alt_text":27,"caption":27,"description":27,"media_type":83,"media_details":339,"post":127,"source_url":355},{"width":340,"height":341,"file":342,"filesize":343,"sizes":344,"image_meta":353},320,414,"2021\u002F12\u002Fhype-cycle.png",79823,{"medium":345,"thumbnail":349},{"file":346,"width":340,"height":247,"mime-type":93,"filesize":347,"source_url":348},"hype-cycle-320x208.png",35531,"https:\u002F\u002Fcdn.medicalfuturist.com\u002Fwp-content\u002Fuploads\u002F2021\u002F12\u002Fhype-cycle-320x208.png",{"file":350,"width":258,"height":258,"mime-type":93,"filesize":351,"source_url":352},"hype-cycle-150x150.png",20118,"https:\u002F\u002Fcdn.medicalfuturist.com\u002Fwp-content\u002Fuploads\u002F2021\u002F12\u002Fhype-cycle-150x150.png",{"aperture":116,"credit":27,"camera":27,"caption":27,"created_timestamp":116,"copyright":27,"focal_length":116,"iso":116,"shutter_speed":116,"title":27,"orientation":116,"keywords":354},[],"https:\u002F\u002Fcdn.medicalfuturist.com\u002Fwp-content\u002Fuploads\u002F2021\u002F12\u002Fhype-cycle.png",{"buy_button_text":357,"leanpub_url":358,"preview":359},"Buy it on Leanpub","https:\u002F\u002Fleanpub.com\u002Ftechadoptioncurve-2025",[360,362,364,366],{"image":361},"https:\u002F\u002Fcdn.medicalfuturist.com\u002Fwp-content\u002Fuploads\u002F2021\u002F12\u002FTAC-Cover.png",{"image":363},"https:\u002F\u002Fcdn.medicalfuturist.com\u002Fwp-content\u002Fuploads\u002F2021\u002F12\u002F0427_hype_cycle_carousel.png",{"image":365},"https:\u002F\u002Fcdn.medicalfuturist.com\u002Fwp-content\u002Fuploads\u002F2021\u002F12\u002F0427_hype_cycle_carousel-02.png",{"image":367},"https:\u002F\u002Fcdn.medicalfuturist.com\u002Fwp-content\u002Fuploads\u002F2021\u002F12\u002F0427_hype_cycle_carousel-03.png",{"yoast_wpseo_title":369,"yoast_wpseo_metadesc":370,"yoast_wpseo_canonical":326},"The Technology Adoption Curve Of The Top 50 Emerging Digital Health Trends - The Medical Futurist","Digital Health Hype Cycle: The future of healthcare is shaping up in front of our eyes. In this book, we analyze the Top 50 Digital Health Trends.",{"self":372,"collection":377,"about":380,"author":383,"replies":385,"wp:featuredmedia":388,"wp:attachment":391,"wp:term":394,"curies":397},[373],{"href":374,"targetHints":375},"https:\u002F\u002Fapi.medicalfuturist.com\u002Fwp-json\u002Fwp\u002Fv2\u002Fbook\u002F37033",{"allow":376},[136],[378],{"href":379},"https:\u002F\u002Fapi.medicalfuturist.com\u002Fwp-json\u002Fwp\u002Fv2\u002Fbook",[381],{"href":382},"https:\u002F\u002Fapi.medicalfuturist.com\u002Fwp-json\u002Fwp\u002Fv2\u002Ftypes\u002Fbook",[384],{"embeddable":26,"href":145},[386],{"embeddable":26,"href":387},"https:\u002F\u002Fapi.medicalfuturist.com\u002Fwp-json\u002Fwp\u002Fv2\u002Fcomments?post=37033",[389],{"embeddable":26,"href":390},"https:\u002F\u002Fapi.medicalfuturist.com\u002Fwp-json\u002Fwp\u002Fv2\u002Fmedia\u002F49955",[392],{"href":393},"https:\u002F\u002Fapi.medicalfuturist.com\u002Fwp-json\u002Fwp\u002Fv2\u002Fmedia?parent=37033",[395],{"taxonomy":177,"embeddable":26,"href":396},"https:\u002F\u002Fapi.medicalfuturist.com\u002Fwp-json\u002Fwp\u002Fv2\u002Fyst_prominent_words?post=37033",[398],{"name":181,"href":182,"templated":26},1789237931961]