Smart multidosing
A method and algorithm for dosing single and multiple drugs for an individual patient includes the step of collecting data from the individual patient. The data includes drugs to be taken by the patient, the patient's age, medical condition, metabolism, and other factors. The method further includes the steps of analyzing the individual patient data in view of dosing criteria established based on patient population data. Based on the analysis, a dose of the single or multiple drugs is recommended. Alternative treatment options can also be recommended.
1. A computer-implemented method for dosing single and multiple drugs for an individual patient, by
a healthcare professional collecting and inputting data from the individual patient including drugs to be taken by the patient into a database with a central artificial intelligence (AI) system stored on computer readable media,
the central AI system analyzing the individual patient data in view of dosing criteria established based on outside patient data from clinical trial databases, electronic medical records, pharmaceutical companies, private databases, and contract research organizations (CROs), wherein outside patients experienced safety and efficacy,
the central AI system creating features from data points of outside patient data variables that affect drug metabolism, the central AI system extracting all the features and creating a model using the features from the outside patient data variables relating dosing to patient condition of the outside patients and effect of drugs on the condition that affect efficacy and toxicity of all drugs taken by the individual patient, the variables including age of patient, weight of patient, known side effects of drugs alone and in combinations with other drugs with the outside patients, known toxicity range as related to median effective dose (ED 50) and dose response points, efficacy ranges, and chronic treatment effect versus acute treatment with outside patients, the central AI system identifying nearest neighbors of the outside patients having similar patient data and/or underwent a treatment plan with similar drug combinations to the individual patient and identifying related clinical trial data with a K-Nearest Neighbor algorithm, the central AI system employing a combination of artificial intelligence techniques, both supervised and unsupervised, including model logic of classifiers and expert rules that are prepopulated by practitioners and published research and are assigned a degree of truth, wherein the central AI system is unique to given inputs and is trained on demand to emphasize individuality of the individual patient and symptoms, the central AI system assigning confidence and weight to the individual patient data, the related clinical trial data, and the outside patient data variables, wherein patient data includes continuous values and discrete values, the central AI system using the classifiers and the expert rules implemented in series, the central AI system comparing the individual patient data to neighboring patient data with weighting schemes, the central AI system calculating the classifiers and model over the range of all doses considered by the model for the individual patient, wherein for each dosage, classifications with confidence intervals are calculated, the model is ran using the dosages mapped out to the classifiers, the outputs from the classifiers and the model are weighted and combined to determine an optimal dose for each drug, and
the central AI system determining a dose or doses of the single or multiple drugs, respectively, for each drug taken by the individual patient and maximizing therapeutic effect while minimizing adverse effects for the combination of drugs taken, the central AI system producing clinical outputs of a recommended dosing range of the single or multiple drugs and in real time communication with a dispensing device to administer the drugs, and displaying the dose in a readable report for a practitioner.
2. The computer-implemented method of claim 1 wherein the individual patient data and the outside patient data is chosen from the group consisting of pharmacokinetics, distribution, prior toxicity and efficacy determinations, age, metabolism, and combinations thereof.
3. The computer-implemented method of claim 1 further including the step of dispensing the drugs to the individual patient in the determined doses.
4. A non-transitory computer-readable media including a computer-implemented algorithm for dosing single and multiple drugs for an individual patient, by
collecting and inputting, data from the individual patient including drugs to be taken by the patient into a database with a central artificial intelligence (AI) system stored on non-transitory computer readable media,
the central AI system analyzing the individual patient data using supervised and unsupervised AI techniques, in view of dosing criteria established based on outside patient data of from clinical trial data from outside databases of clinics, electronic medical records, pharmaceutical companies, private databases, and contract research organizations (CROs), wherein outside patients experienced safety and efficacy,
the central AI system creating features from data points of outside patient data variables that affect drug metabolism, the central AI system extracting all the features and creating a model using the features from the outside patient data relating dosing to patient condition of the outside patients and effect of drugs on the condition that affect efficacy and toxicity of all drugs taken by the individual patient, wherein the variables include age of patient, weight of patient, known side effects of drugs alone and in combinations with other drugs with the outside patients, known toxicity range as related to median effective dose (ED 50) and dose response points, efficacy ranges, and chronic treatment effect versus acute treatment, the central AI system identifying nearest neighbors of the outside patients having similar patient data and/or underwent/undergoing a treatment plan with similar drug combinations to the individual patient and identifying related clinical trial data with a K-Nearest Neighbor algorithm, the central AI system employing the supervised and unsupervised AI techniques, including model logic of classifiers and expert rules that are prepopulated by practitioners and published research and are assigned a degree of truth, wherein the central AI system is unique to given inputs and is trained on demand to emphasize the individuality of the individual patient and symptoms, the central AI system assigning confidence and weight to the individual patient data, the related clinical trial data, and the outside patient data variables, wherein patient data includes continuous values and discrete values, the central AI system using the classifiers and the expert rules implemented in series, the central AI system comparing the individual patient data to neighboring patient data with weighting schemes, the central AI system calculating the classifiers and model over the range of all doses considered by the model for the individual patient, wherein for each dosage, classifications with confidence intervals are calculated, the model is ran using the dosages mapped out to the classifiers, the outputs from the classifiers and the model are weighted and combined to determine an optimal dose for each drug, and
the central AI system determining a dose or doses of the single or multiple drugs, respectively, for each drug taken by the individual patient and maximizing therapeutic effect while minimizing adverse effects for the combination of drugs taken, the central AI system producing clinical outputs of a recommended dosing range of the single or multiple drugs in real time communication with a dispensing device to administer the drugs, and displaying the dose in a readable report for a practitioner.
5. The non-transitory computer-readable media including the computer-implemented algorithm of claim 4 wherein information flows from data input from the individual patient and the outside patient data to the central AI to and from a healthcare professional.
6. A computer-implemented logic engine for dosing multiple drugs, including an algorithm stored on non-transitory computer readable media for collecting outside data from outside databases of clinics, electronic medical records, pharmaceutical companies, private databases, and contract research organizations (CROs) to establish criteria for safely and efficaciously dosing multiple drugs to an individual patient and storing the outside data and individual patient data in a database, analyzing the patient data in view of criteria established from the clinical trial data with a central artificial intelligence (AI) system, the central AI system creating features from data points of outside patient data variables that affect drug metabolism, the central AI system extracting all the features and creating a model using the features from the outside patient data relating dosing to patient condition and effect of drugs on the condition affecting efficacy and toxicity of all drugs taken by the individual patient, wherein the variables include age of patient, weight of patient, known side effects of drugs alone and in combinations with other drugs with the outside patients, known toxicity range as related to median effective dose (ED 50) and dose response points, efficacy ranges, and chronic treatment effect versus acute treatment with outside patients, the central AI system identifying nearest neighbors of the outside patients having similar patient data and/or underwent/undergoing a treatment plan with similar drug combinations to the individual patient and identifying related clinical trial data with a K-Nearest Neighbor algorithm, the central AI system employing a combination of artificial intelligence techniques, both supervised and unsupervised, including model logic of classifiers and expert rules that are prepopulated by practitioners and published research and are assigned a degree of truth, wherein said logic engine is unique to given inputs and is trained on demand to emphasize the individuality of the individual patient and symptoms, the central AI system assigning confidence and weight to the individual patient data, the related clinical trial data, and the outside patient data variables, wherein patient data includes continuous values and discrete values, the central AI system using the classifiers and the expert rules implemented in series, the central AI system comparing the individual patient data to neighboring patient data with weighting schemes, the central AI system calculating the classifiers and model over the range of all doses considered by the model for the individual patient, wherein for each dosage, classifications with confidence intervals are calculated, the model is ran using the dosages mapped out to the classifiers, the outputs from the classifiers and the model are weighted and combined to determine an optimal dose for each drug, and determining a dose or doses of the single or multiple drugs, respectively, for each drug taken, maximizing therapeutic effect while minimizing adverse effects for the combination of drugs taken, producing clinical outputs of a recommended dosing range of the single or multiple drugs in real time communication with a dispensing device to administer the drugs, providing an output in the form of a practitioner readable report.
7. The computer-implemented logic engine of claim 6 wherein the individual patient data includes images chosen from the group consisting of computerized axial tomography (CAT) scans, CT scans, X-rays, magnetic resonance imaging (MRI), ultrasounds, and positron emission tomography (PET) scans.
8. The computer-implemented logic engine of claim 7 wherein the algorithm further performs the steps of analyzing reports from the images, encoding the reports using a natural language processing (NLP) algorithm to extract features, and informing the database of the features, along with structured findings from a radiologist.