Explore chapters and articles related to this topic
The logic in modern medicine: Reasoning and underlying concepts
Published in Milos Jenicek, Foundations of Evidence-Based Medicine, 2019
Both deductive and inductive logic are equally important for medicine. For example, we may particularly need deductive logic (deductively valid arguments) in making crucial therapeutic decisions and handling critical cases and situations in medicine and surgery. On the other hand, there is no meaningful prognosis without the best possible inductive logic and its inductively strong arguments.
Glossary
Published in Pat Croskerry, Karen S. Cosby, Mark L. Graber, Hardeep Singh, Diagnosis, 2017
Pat Croskerry, Karen S. Cosby, Mark L. Graber, Hardeep Singh
inductive logic: a form of logic that begins with specifics and tries to match it to a more general category or draw a more general conclusion; an example is beginning with a symptom and reaching a conclusion about the cause. It has been described as the logic of personal experience. Unlike deductive reasoning, induction always has some degree of uncertainty.
Causality
Published in Stanley Berent, James W. Albers, Neurobehavioral Toxicology, 2012
Stanley Berent, James W. Albers
Inductive logic can be contrasted with deductive reasoning, the former referring often to a combination of empirical inquiry and logic (Burks, 1977). The logic of empirical enquiry follows certain rules that, for the present purpose, can be thought of as scientific method (see Chapter 3). At the risk of oversimplification, one of these rules is to formulate a question and then to test its truth or falsity through systematic observation. Through observation, we discover new truths, some of which might contradict underlying premises we held to be true and some that might go beyond the then present state of knowledge. That is, we might need to change the way in which we previously thought about a given topic, or we might discover a fact that is without causal explanation. Consider for a moment the proposition ‘if A = B and B = C, then A = C’. In keeping with our earlier illustration, let ‘A’ represent exposure to OP pesticides, ‘B’ the symptoms of such exposure, and ‘C’ a person known to have been exposed to an OP compound. As was done by the ‘expert’ in our earlier example, it might appear that one could reasonably deduce from this proposition that person ‘C’ will evidence the symptoms represented by ‘B’ in our model; however, actual observation might reveal that ‘C’ is symptom free. Further, even if person in ‘C’ were found to have such symptoms, we would still not be able to conclude that these were caused by ‘A’, since the dysfunction might be due to something else entirely. This scenario underscores the importance of empirical observation and the need to consider observable phenomena in making logical conclusions. It is rendered inductive through the inclusion of a conditional proposition (a hypothesis, see Chapter 3). If ‘C’ is (or has been) exposed to ‘A’, then ‘C’ will evidence the symptoms described in ‘B’. As will be readily apparent, the hypothesis as described is necessary to conclude objectively an association between ‘A’ and ‘C’, but it is not likely sufficient in itself to conclude a causal relationship. Depending on the nature of the results from observation (e.g., the outcome of a clinical evaluation), the findings may either support or fail to support a positive connection between ‘A’ and ‘C’. Through a series of such propositions and tests, however, evidence can be accumulated that will allow for a properly reasoned, if probabilistic, conclusion about the proposed relationship.
Machine learning techniques applied to the drug design and discovery of new antivirals: a brief look over the past decade
Published in Expert Opinion on Drug Discovery, 2021
Mateus Sá Magalhães Serafim, Valtair Severino dos Santos Júnior, Jadson Castro Gertrudes, Vinícius Gonçalves Maltarollo, Kathia Maria Honorio
Likewise, Langeder et al. (2020) analyzed a large set of natural products and FDA-approved drugs with anti-influenza and anti-rhinovirus activity via PCA, and established a relationship between 15 key properties and the biological activities [126]. Furthermore, Kumar et al. (2021) explored SAR of a chemical space of potential and reported antiviral compounds, where 433 presented anti-coronavirus activity against SARS-CoV and MERS-CoV. After several analyses, including a PCA based on six drug-likeness properties, the authors suggested that compounds containing indole and pyrrolidine scaffolds had the highest biological activity in the dataset, indicating interesting fragments to be explored in potential anti-SARS-CoV-2 agents. The authors also predicted 11 drugs as a possible repurposing strategy to treat COVID-19. For this, they employed a ML tool, Max Chemistry Assistant (DCA [127]), with an inductive logic programming approach, which generates hypotheses that best corroborate the existing structural knowledge of compounds and activity within a given data [128].
Physical therapist’s clinical reasoning in patients with gait impairments from hemiplegia
Published in Physiotherapy Theory and Practice, 2020
A proposed model of diagnostic reasoning is based on the idea of duel-process theory, whereby there are 2 systems of decision making: intuitive and analytical (Croskerry, 2009). The intuitive approach relies heavily on the experience of the decision maker, and the decision maker uses reasoning based on inductive logic. Experienced decision makers will be able to recognize overall patterns (gestalt effects) in the information that is gathered. With the intuitive approach, decisions may be made with uncertainty and based on instinctive first impressions. Alternatively, the analytical approach often occurs under more ideal conditions, with fewer boundaries and great resources, and therefore with less uncertainty. This approach involves hypothesis testing and deductive reasoning, involves critical thinking, and is logically sound. Research in decision making has suggested that robust decision making is more analytical than intuitive. Analytical decision making is based on a systematic approach to remove uncertainty (Croskerry, 2009). This concept was certainly reflected in the findings of this study. Clinicians in the novice group appeared more disorganized in their thought processes and less often described a systematic approach to gait analysis or management. Their verbal and non-verbal responses reflected less certainty in their responses. Those in the expert sample more often described their gait analysis processes in greater detail and these processes reflected a systematic approach. Their answers reflected more certainty and confidence.
Closed-loop discovery platform integration is needed for artificial intelligence to make an impact in drug discovery
Published in Expert Opinion on Drug Discovery, 2019
Semion K. Saikin, Christoph Kreisbeck, Dennis Sheberla, Jill S. Becker, Alán Aspuru-Guzik
The idea of applying ML methods to drug discovery has been around for several decades. In the 1990s, neural networks, inductive logic programming, and support vector machines were employed. At that times, the studies focused on QSAR, target identification, and feature extraction had demonstrated good correlations of the predicted results with the available data [6]. However, the prediction capabilities were limited due to limited computational capabilities and sparse experimental data. These challenges have been ameliorated with the recent advances in computing hardware as well as the increased amount of data from HTS. Major developments in deep neural networks, generative models, and their applications to text and image processing renewed the interest of the scientific community as well as pharma companies to use ML for drug discovery [7–10]. Training these networks, usually, requires a large amount of data, either experimental or computed using microscopic theoretical models.