FlavorDB: a database of flavor molecules.

Flavor is an expression of olfactory and gustatory sensations experienced through a multitude of chemical processes triggered by molecules. Beyond their key role in defining taste and smell, flavor molecules also regulate metabolic processes with consequences to health. Such molecules present in natural sources have been an integral part of human history with limited success in attempts to create synthetic alternatives. Given their utility in various spheres of life such as food and fragrances, it is valuable to have a repository of flavor molecules, their natural sources, physicochemical properties, and sensory responses. FlavorDB (http://cosylab.iiitd.edu.in/flavordb) comprises of 25,595 flavor molecules representing an array of tastes and odors. Among these 2254 molecules are associated with 936 natural ingredients belonging to 34 categories. The dynamic, user-friendly interface of the resource facilitates exploration of flavor molecules for divergent applications: finding molecules matching a desired flavor or structure; exploring molecules of an ingredient; discovering novel food pairings; finding the molecular essence of food ingredients; associating chemical features with a flavor and more. Data-driven studies based on FlavorDB can pave the way for an improved understanding of flavor mechanisms.

Safe and reliable transport of prediction models to new healthcare settings without the need to collect new labeled data.

How can practitioners and clinicians know if a prediction model trained at a different institution can be safely used on their patient population? There is a large body of evidence showing that small changes in the distribution of the covariates used by prediction models may cause them to fail when deployed to new settings. This specific kind of dataset shift, known as covariate shift, is a central challenge to implementing existing prediction models in new healthcare environments. One solution is to collect additional labels in the target population and then fine tune the prediction model to adapt it to the characteristics of the new healthcare setting, which is often referred to as localization. However, collecting new labels can be expensive and time-consuming. To address these issues, we recast the core problem of model transportation in terms of uncertainty quantification, which allows one to know when a model trained in one setting may be safely used in a new healthcare environment of interest. Using methods from conformal prediction, we show how to transport models safely between different settings in the presence of covariate shift, even when all one has access to are covariates from the new setting of interest (e.g. no new labels). Using this approach, the model returns a prediction set that quantifies its uncertainty and is guaranteed to contain the correct label with a user-specified probability (e.g. 90%), a property that is also known as coverage. We show that a weighted conformal inference procedure based on density ratio estimation between the source and target populations can produce prediction sets with the correct level of coverage on real-world data. This allows users to know if a model's predictions can be trusted on their population without the need to collect new labeled data.

The Diagnostic and Triage Accuracy of the GPT-3 Artificial Intelligence Model.

Importance: Artificial intelligence (AI) applications in health care have been effective in many areas of medicine, but they are often trained for a single task using labeled data, making deployment and generalizability challenging. Whether a general-purpose AI language model can perform diagnosis and triage is unknown. Objective: Compare the general-purpose Generative Pre-trained Transformer 3 (GPT-3) AI model's diagnostic and triage performance to attending physicians and lay adults who use the Internet. Design: We compared the accuracy of GPT-3's diagnostic and triage ability for 48 validated case vignettes of both common (e.g., viral illness) and severe (e.g., heart attack) conditions to lay people and practicing physicians. Finally, we examined how well calibrated GPT-3's confidence was for diagnosis and triage. Setting and Participants: The GPT-3 model, a nationally representative sample of lay people, and practicing physicians. Exposure: Validated case vignettes (<60 words; <6th grade reading level). Main Outcomes and Measures: Correct diagnosis, correct triage. Results: Among all cases, GPT-3 replied with the correct diagnosis in its top 3 for 88% (95% CI, 75% to 94%) of cases, compared to 54% (95% CI, 53% to 55%) for lay individuals (p<0.001) and 96% (95% CI, 94% to 97%) for physicians (p=0.0354). GPT-3 triaged (71% correct; 95% CI, 57% to 82%) similarly to lay individuals (74%; 95% CI, 73% to 75%; p=0.73); both were significantly worse than physicians (91%; 95% CI, 89% to 93%; p<0.001). As measured by the Brier score, GPT-3 confidence in its top prediction was reasonably well-calibrated for diagnosis (Brier score = 0.18) and triage (Brier score = 0.22). Conclusions and Relevance: A general-purpose AI language model without any content-specific training could perform diagnosis at levels close to, but below physicians and better than lay individuals. The model was performed less well on triage, where its performance was closer to that of lay individuals.

RecipeDB: a resource for exploring recipes.

Cooking is the act of turning nature into the culture, which has enabled the advent of the omnivorous human diet. The cultural wisdom of processing raw ingredients into delicious dishes is embodied in their cuisines. Recipes thus are the cultural capsules that encode elaborate cooking protocols for evoking sensory satiation as well as providing nourishment. As we stand on the verge of an epidemic of diet-linked disorders, it is eminently important to investigate the culinary correlates of recipes to probe their association with sensory responses as well as consequences for nutrition and health. RecipeDB (https://cosylab.iiitd.edu.in/recipedb) is a structured compilation of recipes, ingredients and nutrition profiles interlinked with flavor profiles and health associations. The repertoire comprises of meticulous integration of 118 171 recipes from cuisines across the globe (6 continents, 26 geocultural regions and 74 countries), cooked using 268 processes (heat, cook, boil, simmer, bake, etc.), by blending over 20 262 diverse ingredients, which are further linked to their flavor molecules (FlavorDB), nutritional profiles (US Department of Agriculture) and empirical records of disease associations obtained from MEDLINE (DietRx). This resource is aimed at facilitating scientific explorations of the culinary space (recipe, ingredient, cooking processes/techniques, dietary styles, etc.) linked to taste (flavor profile) and health (nutrition and disease associations) attributes seeking for divergent applications. Database URL:  https://cosylab.iiitd.edu.in/recipedb.

BitterSweet: Building machine learning models for predicting the bitter and sweet taste of small molecules.

The dichotomy of sweet and bitter tastes is a salient evolutionary feature of human gustatory system with an innate attraction to sweet taste and aversion to bitterness. A better understanding of molecular correlates of bitter-sweet taste gradient is crucial for identification of natural as well as synthetic compounds of desirable taste on this axis. While previous studies have advanced our understanding of the molecular basis of bitter-sweet taste and contributed models for their identification, there is ample scope to enhance these models by meticulous compilation of bitter-sweet molecules and utilization of a wide spectrum of molecular descriptors. Towards these goals, our study provides a structured compilation of bitter, sweet and tasteless molecules and state-of-the-art machine learning models for bitter-sweet taste prediction (BitterSweet). We compare different sets of molecular descriptors for their predictive performance and further identify important features as well as feature blocks. The utility of BitterSweet models is demonstrated by taste prediction on large specialized chemical sets such as FlavorDB, FooDB, SuperSweet, Super Natural II, DSSTox, and DrugBank. To facilitate future research in this direction, we make all datasets and BitterSweet models publicly available, and present an end-to-end software for bitter-sweet taste prediction based on freely available chemical descriptors.

Data-driven analysis of biomedical literature suggests broad-spectrum benefits of culinary herbs and spices.

Spices and herbs are key dietary ingredients used across cultures worldwide. Beyond their use as flavoring and coloring agents, the popularity of these aromatic plant products in culinary preparations has been attributed to their antimicrobial properties. Last few decades have witnessed an exponential growth of biomedical literature investigating the impact of spices and herbs on health, presenting an opportunity to mine for patterns from empirical evidence. Systematic investigation of empirical evidence to enumerate the health consequences of culinary herbs and spices can provide valuable insights into their therapeutic utility. We implemented a text mining protocol to assess the health impact of spices by assimilating, both, their positive and negative effects. We conclude that spices show broad-spectrum benevolence across a range of disease categories in contrast to negative effects that are comparatively narrow-spectrum. We also implement a strategy for disease-specific culinary recommendations of spices based on their therapeutic tradeoff against adverse effects. Further by integrating spice-phytochemical-disease associations, we identify bioactive spice phytochemicals potentially involved in their therapeutic effects. Our study provides a systems perspective on health effects of culinary spices and herbs with applications for dietary recommendations as well as identification of phytochemicals potentially involved in underlying molecular mechanisms.