Translation of medical evidence into computational evidence and applications thereof

Abstract

A computational evidence platform extracts clinical concepts from medical evidence sources and creates a database of elemental diagnostic factors and elemental investigations links to medical conditions. Input from a person groups factors and investigations makes corrections and adds a ranking. Elemental factors and investigations do not include information specific to their associated conditions but include synonyms and a link to a medical ontology. A patient state is determined by extracting patient known diagnostic factors and investigation results from the patient chart. These known factors and results are matched to the database and a ranking of likely conditions are output. Next-best actions per condition are output by determining factors not yet known and investigations not yet performed. Next-best actions across conditions are determined by performing a recursive tree search of the database and assuming that unknown factors are now known to generate a score for each assumption.

We claim: 
     
         1 . A method of displaying medical conditions linked to a diagnostic factor, said method comprising:
 inputting medical evidence as text that includes a medical condition and at least one diagnostic factor caused by said medical condition;   transforming said textual medical evidence into at least diagnostic factor data structure, said diagnostic factor data structure including a name, at least one synonym of said diagnostic factor and a unique identifier to a link in medical oncology identifying said diagnostic factor; and   creating a graph database having at least said diagnostic factor data structure as a node and a plurality of nodes each representing medical conditions, each of said medical condition nodes having a link to said diagnostic factor node indicating that said each medical condition causes said diagnostic factor.   
     
     
         2 . A method as recited in  claim 1  further comprising:
 displaying said graph database with said diagnostic factor node and said plurality of medical condition nodes. 
 
     
     
         3 . A method as recited in  claim 1  further comprising:
 creating and displaying a differential diagnosis link between a first one of said medical condition nodes to a second one of said medical condition nodes indicating that said second medical condition is a differential diagnosis of said first medical condition. 
 
     
     
         4 . A method as recited in  claim 1  wherein said diagnostic factor data structure further includes a temporality attribute indicating whether or not a patient is currently exhibiting said diagnostic factor, a negation attribute indicating whether not said diagnostic factor is present in said patient, and an experience attribute indicating whether not said patient experiences said diagnostic factor personally. 
     
     
         5 . A method of updating a citation for medical evidence, said method comprising:
 inputting a portion of electronic text pertaining to a medical condition from a medical evidence source;   processing said electronic text using natural language processing (NLP) to produce a plurality of diagnostic factors indicative of said condition;   extracting a citation from said electronic text that supports said diagnostic factors being indicative of said condition;   calculating a hash value of said citation;   storing said condition and each of said diagnostic factors as a separate entry in a database, said condition linked to each of said diagnostic factors; and   storing said hash value of said citation as an entry in said database in association with said stored diagnostic factor entries.   
     
     
         6 . A method as recited in  claim 5  further comprising:
 after said steps of storing, automatically extracting said citation from said electronic text and calculating a new hash value of said citation; and 
 displaying a warning on a computing device when it is determined that said new hash value does not match said hash value. 
 
     
     
         7 . A method as recited in  claim 5  wherein an entry of each of said diagnostic factors in said database does not include information specific to said condition. 
     
     
         8 . A method of displaying a citation for a medical condition, said method comprising:
 displaying on a computing device a likelihood that a patient has a medical condition along with a plurality of known diagnostic factors indicating that said patient has said medical condition;   receiving an input selecting said medical condition;   displaying a graph database showing said medical condition linked to said diagnostic factors;   receiving a selection of a group of at least one of said diagnostic factors;   displaying said group along with a citation link that supports said group indicating said medical condition;   receiving a selection of said citation link; and   displaying said citation including a link to a reference supporting said citation.   
     
     
         9 . A method of determining next-best actions across medical conditions, said method comprising:
 identifying all unknown diagnostic factor entries in a database that each represent a diagnostic factor not known to be present or absent in a patient, each unknown diagnostic factor entry being linked to a medical condition entry in said database representing a medical condition but not including information specific to said each medical condition;   executing a recursive search algorithm and performing the following   
       choosing a next unknown diagnostic factor entry in said database and assuming it is now known to be present or absent in said patient,
 generating a ranking for each potential medical condition of said patient based on known diagnostic factors entries in said database that each represent a diagnostic factor known to be present or absent in said patient, and 
 scoring said rankings of said each potential medical conditions to generate a score for said chosen next unknown diagnostic factor entry, 
 continuing to perform said executing step recursively until a terminating condition is met; and 
 when said terminating condition is met, displaying a list of unknown diagnostic factors corresponding to said chosen next unknown diagnostic factor entries in order of the score for each of said chosen next unknown diagnostic factor entries. 
 
     
     
         10 . A method as recited in  claim 9  wherein said ranking for said each potential medical condition is based upon a rank value of each of said known diagnostic factor entries that is stored in an entry in between said each known diagnostic factor and its corresponding medical condition, said rank value not being stored in said known diagnostic factor entry. 
     
     
         11 . A method as recited in  claim 9  wherein said score for said next unknown diagnostic factor entry is based upon a dollar cost of querying said unknown diagnostic factor, a time of querying said unknown diagnostic factor and a risk of querying said unknown diagnostic factor. 
     
     
         12 . A method as recited in  claim 9  wherein said search algorithm is a tree search algorithm. 
     
     
         13 . A method as recited in  claim 12  wherein said tree search algorithm is the Monte Carlo tree search algorithm. 
     
     
         14 . A method as recited in  claim 9  wherein said score for said chosen next unknown diagnostic factor entry indicates a difference between said rankings of said each medical conditions. 
     
     
         15 . A method as recited in  claim 9  wherein each said known and unknown diagnostic factor entry includes at least one synonym of said known and unknown diagnostic factor. 
     
     
         16 . A method as recited in  claim 9  wherein each said known and unknown diagnostic factor entry includes a temporality attribute, a negation attribute or an experiencer attribute. 
     
     
         17 . A method as recited in  claim 9  wherein each of said known and unknown diagnostic factor entries does not include any information specific to the medical condition entry linked to said each known and unknown diagnostic factor entries.